ARCore: Augmented reality at Android scale

Posted by Dave Burke, VP, Android Engineering

With more than two billion active devices, Android is the largest mobile platform in the world. And for the past nine years, we've worked to create a rich set of tools, frameworks and APIs that deliver developers' creations to people everywhere. Today, we're releasing a preview of a new software development kit (SDK) called ARCore. It brings augmented reality capabilities to existing and future Android phones. Developers can start experimenting with it right now.

We've been developing the fundamental technologies that power mobile AR over the last three years with Tango, and ARCore is built on that work. But, it works without any additional hardware, which means it can scale across the Android ecosystem. ARCore will run on millions of devices, starting today with the Pixel and Samsung's S8, running 7.0 Nougat and above. We're targeting 100 million devices at the end of the preview. We're working with manufacturers like Samsung, Huawei, LG, ASUS and others to make this possible with a consistent bar for quality and high performance.

ARCore works with Java/OpenGL, Unity and Unreal and focuses on three things:

• Motion tracking: Using the phone's camera to observe feature points in the room and IMU sensor data, ARCore determines both the position and orientation (pose) of the phone as it moves. Virtual objects remain accurately placed.
• Environmental understanding: It's common for AR objects to be placed on a floor or a table. ARCore can detect horizontal surfaces using the same feature points it uses for motion tracking.
• Light estimation: ARCore observes the ambient light in the environment and makes it possible for developers to light virtual objects in ways that match their surroundings, making their appearance even more realistic.

Alongside ARCore, we've been investing in apps and services which will further support developers in creating great AR experiences. We built Blocks and Tilt Brush to make it easy for anyone to quickly create great 3D content for use in AR apps. As we mentioned at I/O, we're also working on Visual Positioning Service (VPS), a service which will enable world scale AR experiences well beyond a tabletop. And we think the Web will be a critical component of the future of AR, so we're also releasing prototype browsers for web developers so they can start experimenting with AR, too. These custom browsers allow developers to create AR-enhanced websites and run them on both Android/ARCore and iOS/ARKit.

ARCore is our next step in bringing AR to everyone, and we'll have more to share later this year. Let us know what you think through GitHub, and check out our new AR Experiments showcase where you can find some fun examples of what's possible. Show us what you build on social media with #ARCore; we'll be resharing some of our favorites.

Read the full article here by Android Developers Blog

L’etichetta Fact Check da oggi disponibile in tutti i paesi nella ricerca Google e in Google News

Google è stata creata con l’obiettivo di aiutare gli utenti a trovare informazioni utili, offrendo visibilità ai contenuti che gli editori creano.

Tuttavia, con migliaia di nuovi articoli pubblicati online ogni minuto di ogni giorno, la quantità di contenuti con cui si confrontano gli utenti può risultare eccessiva. E purtroppo, non tutti questi contenuti sono aderenti ai fatti o veri, rendendo così difficile per i lettori distinguere i fatti da ciò che è falso. Ecco perché ad ottobre, insieme ai nostri partner di Jigsaw, abbiamo annunciato che in alcuni Paesi avremmo iniziato a consentire agli editori di mostrare l’etichetta "Fact Check" in Google News. Questa etichetta consente di identificare in modo più immediato gli articoli di verifica dei fatti.

Dopo aver valutato i riscontri ricevuti da parte degli utenti e degli editori, abbiamo deciso di rendere disponibile l’etichetta Fact Check in Google News ovunque e di estenderla al motore di ricerca, a livello globale e in tutte le lingue. Per la prima volta, quando viene effettuata una ricerca su Google che restituisce un risultato che contiene la verifica dei fatti di uno o più affermazioni pubbliche, questa informazione verrà chiaramente visualizzata nella pagina dei risultati di ricerca. Lo snippet mostrerà informazioni sulla dichiarazione verificata, da chi è stata fatta e se una fonte ha verificato quella particolare dichiarazione.

Queste informazioni non saranno disponibili per qualsiasi risultato e potrebbero esserci pagine di risultati di ricerca in cui diverse fonti hanno verificato la stessa affermazione raggiungendo però conclusioni diverse. Queste verifiche dei fatti naturalmente non sono effettuate da Google e potremmo anche non essere d'accordo con i risultati, proprio come diversi articoli di fact checking potrebbero essere in disaccordo tra loro, tuttavia riteniamo che sia utile per le persone capire il grado di consenso attorno a un argomento e avere informazioni chiare su quali fonti concordano. Rendendo queste attività di fact checking più visibili nei risultati di ricerca, riteniamo che gli utenti possano esaminarle e valutarle con maggiore facilità per formarsi così opinioni e pareri informati.

Per poter usufruire di questa etichetta, gli editori devono utilizzare il markup ClaimReview di Schema.org sulle pagine nelle quali effettuano il fact checking di dichiarazioni pubbliche (informazioni maggiori qui) o usare il widget Share the Facts sviluppato dal Duke University Reporters Lab e Jigsaw. Solo gli editori che sono algoritmicamente determinati come fonte autorevole di informazioni si qualificheranno per essere inclusi. Infine, i contenuti dovranno rispettare le norme generali che si applicano a tutti i tag di dati strutturati e ai criteri di Google News Publisher per il fact checking. Se un editore o un articolo di fact checking non raggiunge questi standard o non rispetta tali norme, potremo, a nostra discrezione, ignorare il markup.

Tutto ciò non sarebbe stato possibile senza l'aiuto di altre organizzazioni e senza il sostegno della comunità di fact checking, che è cresciuta fino a includere più di 115 organizzazioni. Se volete saperne di più visitate il nostro Centro assistenza per gli utenti.

Scritto da: Justin Kosslyn (Product Manager, Jigsaw) e Cong Yu (Research Scientist, Google Research) Justin Kosslyn, Cong YuProduct Manager, Research ScientistJigsaw, Google Research

Read the full article here by Google Italia Blog

An Upgrade to SyntaxNet, New Models and a Parsing Competition

At Google, we continuously improve the language understanding capabilities used in applications ranging from generation of email responses to translation. Last summer, we open-sourced SyntaxNet, a neural-network framework for analyzing and understanding the grammatical structure of sentences. Included in our release was Parsey McParseface, a state-of-the-art model that we had trained for analyzing English, followed quickly by a collection of pre-trained models for 40 additional languages, which we dubbed Parsey's Cousins. While we were excited to share our research and to provide these resources to the broader community, building machine learning systems that work well for languages other than English remains an ongoing challenge. We are excited to announce a few new research resources, available now, that address this problem.

SyntaxNet Upgrade
We are releasing a major upgrade to SyntaxNet. This upgrade incorporates nearly a year’s worth of our research on multilingual language understanding, and is available to anyone interested in building systems for processing and understanding text. At the core of the upgrade is a new technology that enables learning of richly layered representations of input sentences. More specifically, the upgrade extends TensorFlow to allow joint modeling of multiple levels of linguistic structure, and to allow neural-network architectures to be created dynamically during processing of a sentence or document.

Our upgrade makes it, for example, easy to build character-based models that learn to compose individual characters into words (e.g. ‘c-a-t’ spells ‘cat’). By doing so, the models can learn that words can be related to each other because they share common parts (e.g. ‘cats’ is the plural of ‘cat’ and shares the same stem; ‘wildcat’ is a type of ‘cat’). Parsey and Parsey’s Cousins, on the other hand, operated over sequences of words. As a result, they were forced to memorize words seen during training and relied mostly on the context to determine the grammatical function of previously unseen words.

As an example, consider the following (meaningless but grammatically correct) sentence:

This sentence was originally coined by Andrew Ingraham who explained: “You do not know what this means; nor do I. But if we assume that it is English, we know that the doshes are distimmed by the gostak. We know too that one distimmer of doshes is a gostak." Systematic patterns in morphology and syntax allow us to guess the grammatical function of words even when they are completely novel: we understand that ‘doshes’ is the plural of the noun ‘dosh’ (similar to the ‘cats’ example above) or that ‘distim’ is the third person singular of the verb distim. Based on this analysis we can then derive the overall structure of this sentence even though we have never seen the words before.

ParseySaurus
To showcase the new capabilities provided by our upgrade to SyntaxNet, we are releasing a set of new pretrained models called ParseySaurus. These models use the character-based input representation mentioned above and are thus much better at predicting the meaning of new words based both on their spelling and how they are used in context. The ParseySaurus models are far more accurate than Parsey’s Cousins (reducing errors by as much as 25%), particularly for morphologically-rich languages like Russian, or agglutinative languages like Turkish and Hungarian. In those languages there can be dozens of forms for each word and many of these forms might never be observed during training - even in a very large corpus.

Consider the following fictitious Russian sentence, where again the stems are meaningless, but the suffixes define an unambiguous interpretation of the sentence structure:

Even though our Russian ParseySaurus model has never seen these words, it can correctly analyze the sentence by inspecting the character sequences which constitute each word. In doing so, the system can determine many properties of the words (notice how many more morphological features there are here than in the English example). To see the sentence as ParseySaurus does, here is a visualization of how the model analyzes this sentence:

Each square represents one node in the neural network graph, and lines show the connections between them. The left-side “tail” of the graph shows the model consuming the input as one long string of characters. These are intermittently passed to the right side, where the rich web of connections shows the model composing words into phrases and producing a syntactic parse. Check out the full-size rendering here.

A Competition
You might be wondering whether character-based modeling are all we need or whether there are other techniques that might be important. SyntaxNet has lots more to offer, like beam search and different training objectives, but there are of course also many other possibilities. To find out what works well in practice we are helping co-organize, together with Charles University and other colleagues, a multilingual parsing competition at this year’s Conference on Computational Natural Language Learning (CoNLL) with the goal of building syntactic parsing systems that work well in real-world settings and for 45 different languages.

The competition is made possible by the Universal Dependencies (UD) initiative, whose goal is to develop cross-linguistically consistent treebanks. Because machine learned models can only be as good as the data that they have access to, we have been contributing data to UD since 2013. For the competition, we partnered with UD and DFKI to build a new multilingual evaluation set consisting of 1000 sentences that have been translated into 20+ different languages and annotated by linguists with parse trees. This evaluation set is the first of its kind (in the past, each language had its own independent evaluation set) and will enable more consistent cross-lingual comparisons. Because the sentences have the same meaning and have been annotated according to the same guidelines, we will be able to get closer to answering the question of which languages might be harder to parse.

We hope that the upgraded SyntaxNet framework and our the pre-trained ParseySaurus models will inspire researchers to participate in the competition. We have additionally created a tutorial showing how to load a Docker image and train models on the Google Cloud Platform, to facilitate participation by smaller teams with limited resources. So, if you have an idea for making your own models with the SyntaxNet framework, sign up to compete! We believe that the configurations that we are releasing are a good place to start, but we look forward to seeing how participants will be able to extend and improve these models or perhaps create better ones!

Thanks to everyone involved who made this competition happen, including our collaborators at UD-Pipe, who provide another baseline implementation to make it easy to enter the competition. Happy parsing from the main developers, Chris Alberti, Daniel Andor, Ivan Bogatyy, Mark Omernick, Zora Tung and Ji Ma!

By David Weiss and Slav Petrov, Research Scientists

Read the full article here by Google Open Source Blog

O-MG, the Developer Preview of Android O is here!

Posted by Dave Burke, VP of Engineering

Since the first launch in 2008, the Android project has thrived on the incredible feedback from our vibrant ecosystems of app developers and device makers, as well as of course our users. More recently, we've been pushing hard on improving our engineering processes so we can share our work earlier and more openly with our partners.

So, today, I'm excited to share a first developer preview of the next version of the OS: Android O. The usual caveats apply: it's early days, there are more features coming, and there's still plenty of stabilization and performance work ahead of us. But it's booting :).

Over the course of the next several months, we'll be releasing updated developer previews, and we'll be doing a deep dive on all things Android at Google I/O in May. In the meantime, we'd love your feedback on trying out new features, and of course testing your apps on the new OS.

What's new in O?

Android O introduces a number of new features and APIs to use in your apps. Here's are just a few new things for you to start trying in this first Developer Preview:

Background limits: Building on the work we began in Nougat, Android O puts a big priority on improving a user's battery life and the device's interactive performance. To make this possible, we've put additional automatic limits on what apps can do in the background, in three main areas: implicit broadcasts, background services, and location updates. These changes will make it easier to create apps that have minimal impact on a user's device and battery. Background limits represent a significant change in Android, so we want every developer to get familiar with them. Check out the documentation on background execution limits and background location limits for details.

Notification channels: Android O also introduces notification channels, which are new app-defined categories for notification content. Channels let developers give users fine-grained control over different kinds of notifications — users can block or change the behavior of each channel individually, rather than managing all of the app's notifications together.

Notification channels let users control your app's notification categories

Android O also adds new visuals and grouping to notifications that make it easier for users to see what's going on when they have an incoming message or are glancing at the notification shade.

Autofill APIs: Android users already depend on a range of password managers to autofill login details and repetitive information, which makes setting up new apps or placing transactions easier. Now we are making this work more easily across the ecosystem by adding platform support for autofill. Users can select an autofill app, similar to the way they select a keyboard app. The autofill app stores and secures user data, such as addresses, user names, and even passwords. For apps that want to handle autofill, we're adding new APIs to implement an Autofill service.

PIP for handsets and new windowing features: Picture in Picture (PIP) display is now available on phones and tablets, so users can continue watching a video while they're answering a chat or hailing a car. Apps can put themselves in PiP mode from the resumed or a pausing state where the system supports it - and you can specify the aspect ratio and a set of custom interactions (such as play/pause). Other new windowing features include a new app overlay window for apps to use instead of system alert window, and multi-display support for launching an activity on a remote display.

Font resources in XML: Fonts are now a fully supported resource type in Android O. Apps can now use fonts in XML layouts as well as define font families in XML — declaring the font style and weight along with the font files.

Adaptive icons: To help you integrate better with the device UI, you can now create adaptive icons that the system displays in different shapes, based on a mask selected by the device. The system also animates interactions with the icons, and them in the launcher, shortcuts, Settings, sharing dialogs, and in the overview screen.

Adaptive icons display in a variety of shapes across different device models.

Wide-gamut color for apps: Android developers of imaging apps can now take advantage of new devices that have a wide-gamut color capable display. To display wide gamut images, apps will need to enable a flag in their manifest (per activity) and load bitmaps with an embedded wide color profile (AdobeRGB, Pro Photo RGB, DCI-P3, etc.).

Connectivity: For the ultimate in audio fidelity, Android O now also supports high-quality Bluetooth audio codecs such as LDAC codec. We're also adding new Wi-Fi features as well, like Wi-Fi Aware, previously known as Neighbor Awareness Networking (NAN). On devices with the appropriate hardware, apps and nearby devices can discover and communicate over Wi-Fi without an Internet access point. We're working with our hardware partners to bring Wi-Fi Aware technology to devices as soon as possible.

The Telecom framework is extending ConnectionService APIs to enable third party calling apps integrate with System UI and operate seamlessly with other audio apps. For instance, apps can have their calls displayed and controlled in different kinds of UIs such as car head units.

Keyboard navigation: With the advent of Google Play apps on Chrome OS and other large form factors, we're seeing a resurgence of keyboard navigation use within these apps. In Android O we focused on building a more reliable, predictable model for "arrow" and "tab" navigation that aids both developers and end users.

AAudio API for Pro Audio: AAudio is a new native API that's designed specifically for apps that require high-performance, low-latency audio. Apps using AAudio read and write data via streams. In the Developer Preview we're releasing an early version of this new API to get your feedback.

WebView enhancements: In Android Nougat we introduced an optional multiprocess mode for WebView that moved the handling of web content into an isolated process. In Android O, we're enabling multiprocess mode by default and adding an API to let your app handle errors and crashes, for enhanced security and improved app stability. As a further security measure, you can now opt in your app's WebView objects to verify URLs through Google Safe Browsing.

Java 8 Language APIs and runtime optimizations: Android now supports several new Java Language APIs, including the new java.time API. In addition, the Android Runtime is faster than ever before, with improvements of up to 2x on some application benchmarks.

Partner platform contributions: Hardware manufacturers and silicon partners have accelerated fixes and enhancements to the Android platform in the O release. For example, Sony has contributed more than 30 feature enhancements, including the LDAC codec, and 250 bug fixes to Android O.

Get started in a few simple steps

First, make your app compatible to give your users a seamless transition to Android O. Just download a device system image or emulator system image, install your current app, and test -- the app should run and look great, and handle behavior changes properly. After you've made any necessary updates, we recommend publishing to Google Play right away without changing the app's platform targeting.

Building with Android O

When you're ready, dive in to O in depth to learn about everything you can take advantage of for your app. Visit the O Developer Preview site for details on the preview timeline, behavior changes, new APIs, and support resources.

Plan how your app will support background limits and other changes. Try out some of the great new features in your app -- notification channels, PIP, adaptive icons, font resources in XML, autosizing TextView, and many others. To make it easier to explore the new APIs in Android O, we've brought the API diff report online, along with the Android O API reference.

Coming later today, the latest canary version of Android Studio 2.4 includes new features to help you get started with Android O. When this update is available, you can download and set up the O preview SDK from inside Android Studio, then use Android O's XML font resources and autosizing TextView in the Layout Editor. Watch for more Android O support coming in the weeks ahead.

We're also releasing an alpha version of the 26.0.0 support library for you to try.

Preview updates

The O Developer Preview includes an updated SDK with system images for testing on the official Android Emulator and on Nexus 5X, Nexus 6P, Nexus Player, Pixel, Pixel XL and Pixel C devices. If you're building for wearables, there's also an emulator for testing Android Wear 2.0 on Android O.

We plan to update the preview system images and SDK regularly throughout the O Developer Preview. This initial preview release is for developers only and not intended for daily or consumer use, so we're making it available by manual download and flash only. Downloads and instructions are here.

As we get closer to a final product, we'll be inviting consumers to try it out as well, and we'll open up enrollments through Android Beta at that time. Stay tuned for details, but for now please note that Android Beta is not currently available for Android O.

Give us your feedback

As always, your feedback is crucial, so please let us know what you think — the sooner we hear from you, the more of your feedback we can integrate. When you find issues, please report them here. We've moved to a more robust tool, Issue Tracker, which is also used internally at Google to track bugs and feature requests during product development. We hope you'll find it easier to use.

Read the full article here by Android Developers Blog

100 announcements (!) from Google Cloud Next '17

San Francisco — What a week! Google Cloud Next ‘17 has come to the end, but really, it’s just the beginning. We welcomed

10,000+ attendees including customers, partners, developers, IT leaders, engineers, press, analysts, cloud enthusiasts (and skeptics). Together we engaged in 3 days of keynotes, 200+ sessions, and 4 invitation-only summits. Hard to believe this was our first show as all of Google Cloud with GCP, G Suite, Chrome, Maps and Education. Thank you to all who were here with us in San Francisco this week, and we hope to see you next year.

If you’re a fan of video highlights, we’ve got you covered. Check out our Day 1 keynote (in less than 4 minutes) and Day 2 keynote (in under 5!).

One of the common refrains from customers and partners throughout the conference was “Wow, you’ve been busy. I can’t believe how many announcements you’ve had at Next!” So we decided to count all the announcements from across Google Cloud and in fact we had 100 (!) announcements this week.

For the list lovers amongst you, we’ve compiled a handy-dandy run-down of our announcements from the past few days:

Google Cloud Acquisitions

Google Cloud is excited to welcome two new acquisitions to the Google Cloud family this week, Kaggle and AppBridge.

1. Kaggle - Kaggle is one of the world's largest communities of data scientists and machine learning enthusiasts. Kaggle and Google Cloud will continue to support machine learning training and deployment services in addition to offering the community the ability to store and query large datasets.

2. AppBridge - Google Cloud acquired Vancouver-based AppBridge this week, which helps you migrate data from on-prem file servers into G Suite and Google Drive.

Google Cloud Security

Google Cloud brings a suite of new security features to Google Cloud Platform and G Suite designed to help safeguard your company’s assets and prevent disruption to your business: 

3. Identity-Aware Proxy (IAP) for Google Cloud Platform (Beta) - Identity-Aware Proxy lets you provide access to applications based on risk, rather than using a VPN. It provides secure application access from anywhere, restricts access by user, identity and group, deploys with integrated phishing resistant Security Key and is easier to setup than end-user VPN.

4. Data Loss Prevention (DLP) for Google Cloud Platform (Beta) - Data Loss Prevention API lets you scan data for 40+ sensitive data types, and is used as part of DLP in Gmail and Drive. You can find and redact sensitive data stored in GCP, invigorate old applications with new sensitive data sensing “smarts” and use predefined detectors as well as customize your own.

5. Key Management Service (KMS) for Google Cloud Platform (GA) - Key Management Service allows you to generate, use, rotate, and destroy symmetric encryption keys for use in the cloud.

6. Security Key Enforcement (SKE) for Google Cloud Platform (GA) - Security Key Enforcement allows you to require security keys be used as the 2-Step verification factor for enhanced anti-phishing security whenever a GCP application is accessed.

7. Vault for Google Drive (GA) - Google Vault is the eDiscovery and archiving solution for G Suite. Vault enables admins to easily manage their G Suite data lifecycle and search, preview and export the G Suite data in their domain. Vault for Drive enables full support for Google Drive content, including Team Drive files.

8. Google-designed security chip, Titan - Google uses Titan to establish hardware root of trust, allowing us to securely identify and authenticate legitimate access at the hardware level. Titan includes a hardware random number generator, performs cryptographic operations in the isolated memory, and has a dedicated secure processor (on-chip).

Google Cloud Platform - Data Analytics

New GCP data analytics products and services help organizations solve business problems with data, rather than spending time and resources building, integrating and managing the underlying infrastructure:

9. BigQuery Data Transfer Service (Private Beta) - BigQuery Data Transfer Service makes it easy for users to quickly get value from all their Google-managed advertising datasets. With just a few clicks, marketing analysts can schedule data imports from Google Adwords, DoubleClick Campaign Manager, DoubleClick for Publishers and YouTube Content and Channel Owner reports.

10. Cloud Dataprep (Private Beta) - Cloud Dataprep is a new managed data service, built in collaboration with Trifacta, that makes it faster and easier for BigQuery end-users to visually explore and prepare data for analysis without the need for dedicated data engineer resources.

11. New Commercial Datasets - Businesses often look for datasets (public or commercial) outside their organizational boundaries. Commercial datasets offered include financial market data from Xignite, residential real-estate valuations (historical and projected) from HouseCanary, predictions for when a house will go on sale from Remine, historical weather data from AccuWeather, and news archives from Dow Jones, all immediately ready for use in BigQuery (with more to come as new partners join the program).

12. Python for Google Cloud Dataflow in GA - Cloud Dataflow is a fully managed data processing service supporting both batch and stream execution of pipelines. Until recently, these benefits have been available solely to Java developers. Now there’s a Python SDK for Cloud Dataflow in GA.

13. Stackdriver Monitoring for Cloud Dataflow (Beta) - We’ve integrated Cloud Dataflow with Stackdriver Monitoring so that you can access and analyze Cloud Dataflow job metrics and create alerts for specific Dataflow job conditions.

14. Google Cloud Datalab in GA - This interactive data science workflow tool makes it easy to do iterative model and data analysis in a Jupyter notebook-based environment using standard SQL, Python and shell commands.

15. Cloud Dataproc updates - Our fully managed service for running Apache Spark, Flink and Hadoop pipelines has new support for restarting failed jobs (including automatic restart as needed) in beta, the ability to create single-node clusters for lightweight sandbox development, in beta, GPU support, and the cloud labels feature, for more flexibility managing your Dataproc resources, is now GA.

Google Cloud Platform - Database Services

New GCP databases and database features round out a platform on which developers can build great applications across a spectrum of use cases:

16. Cloud SQL for Postgre SQL (Beta) - Cloud SQL for PostgreSQL implements the same design principles currently reflected in Cloud SQL for MySQL, namely, the ability to securely store and connect to your relational data via open standards.

17. Microsoft SQL Server Enterprise (GA) - Available on Google Compute Engine, plus support for Windows Server Failover Clustering (WSFC) and SQL Server AlwaysOn Availability (GA).

18. Cloud SQL for MySQL improvements - Increased performance for demanding workloads via 32-core instances with up to 208GB of RAM, and central management of resources via Identity and Access Management (IAM) controls.

19. Cloud Spanner - Launched a month ago, but still, it would be remiss not to mention it because, hello, it’s Cloud Spanner! The industry’s first horizontally scalable, globally consistent, relational database service.

20. SSD persistent-disk performance improvements - SSD persistent disks now have increased throughput and IOPS performance, which are particularly beneficial for database and analytics workloads. Read these docs for complete details about persistent-disk performance.

21. Federated query on Cloud Bigtable - We’ve extended BigQuery’s reach to query data inside Cloud Bigtable, the NoSQL database service for massive analytic or operational workloads that require low latency and high throughput (particularly common in Financial Services and IoT use cases).

Google Cloud Platform - Machine Learning Services

New GCP Cloud Machine Learning services bolster our efforts to make machine learning accessible to organizations of all sizes and sophistication:

22.  Cloud Machine Learning Engine (GA) - Cloud ML Engine, now generally available, is for organizations that want to train and deploy their own models into production in the cloud.

23. Cloud Video Intelligence API (Private Beta) - A first of its kind, Cloud Video Intelligence API lets developers easily search and discover video content by providing information about entities (nouns such as “dog,” “flower”, or “human” or verbs such as “run,” “swim,” or “fly”) inside video content.

24. Cloud Vision API (GA) - Cloud Vision API reaches GA and offers new capabilities for enterprises and partners to classify a more diverse set of images. The API can now recognize millions of entities from Google’s Knowledge Graph and offers enhanced OCR capabilities that can extract text from scans of text-heavy documents such as legal contracts or research papers or books.

25. Machine learning Advanced Solution Lab (ASL) - ASL provides dedicated facilities for our customers to directly collaborate with Google’s machine-learning experts to apply ML to their most pressing challenges.

26. Cloud Jobs API - A powerful aid to job search and discovery, Cloud Jobs API now has new features such as Commute Search, which will return relevant jobs based on desired commute time and preferred mode of transportation.

27. Machine Learning Startup Competition - We announced a Machine Learning Startup Competition in collaboration with venture capital firms Data Collective and Emergence Capital, and with additional support from a16z, Greylock Partners, GV, Kleiner Perkins Caufield & Byers and Sequoia Capital.

Google Cloud Platform - Pricing & Support

New GCP pricing continues our intention to create customer-friendly pricing that’s as smart as our products; and support services that are geared towards meeting our customers where they are:

28. Compute Engine price cuts - Continuing our history of pricing leadership, we’ve cut Google Compute Engine prices by up to 8%.

29. Committed Use Discounts - With Committed Use Discounts, customers can receive a discount of up to 57% off our list price, in exchange for a one or three year purchase commitment paid monthly, with no upfront costs.

30. Free trial extended to 12 months - We’ve extended our free trial from 60 days to 12 months, allowing you to use your $300 credit across all GCP services and APIs, at your own pace and schedule. Plus, we’re introduced new Always Free products -- non-expiring usage limits that you can use to test and develop applications at no cost. Visit the Google Cloud Platform Free Tier page for details.

31. Engineering Support - Our new Engineering Support offering is a role-based subscription model that allows us to match engineer to engineer, to meet you where your business is, no matter what stage of development you’re in. It has 3 tiers:

• Development engineering support - ideal for developers or QA engineers that can manage with a response within four to eight business hours, priced at $100/user per month.
  
• Production engineering support provides a one-hour response time for critical issues at $250/user per month.
  
• On-call engineering support pages a Google engineer and delivers a 15-minute response time 24x7 for critical issues at $1,500/user per month.
  

32. http://ift.tt/2msqs6b site - Google Cloud Platform Community is a new site to learn, connect and share with other people like you, who are interested in GCP. You can follow along with tutorials or submit one yourself, find meetups in your area, and learn about community resources for GCP support, open source projects and more.

Google Cloud Platform - Developer Platforms & Tools

New GCP developer platforms and tools reinforce our commitment to openness and choice and giving you what you need to move fast and focus on great code.

33. Google AppEngine Flex (GA) - We announced a major expansion of our popular App Engine platform to new developer communities that emphasizes openness, developer choice, and application portability.

34. Cloud Functions (Beta) - Google Cloud Functions has launched into public beta. It is a serverless environment for creating event-driven applications and microservices, letting you build and connect cloud services with code.

35. Firebase integration with GCP (GA) - Firebase Storage is now Google Cloud Storage for Firebase and adds support for multiple buckets, support for linking to existing buckets, and integrates with Google Cloud Functions.

36. Cloud Container Builder - Cloud Container Builder is standalone tool that lets you build your Docker containers on GCP regardless of deployment environment. It’s a fast, reliable, and consistent way to package your software into containers as part of an automated workflow.

37. Community Tutorials (Beta)  - With community tutorials, anyone can now submit or request a technical how-to for Google Cloud Platform.

Google Cloud Platform - Infrastructure

Secure, global and high-performance, we’ve built our cloud for the long haul. This week we announced a slew of new infrastructure updates.

38. New data center region: California - This new GCP region delivers lower latency for customers on the West Coast of the U.S. and adjacent geographic areas. Like other Google Cloud regions, it will feature a minimum of three zones, benefit from Google’s global, private fibre network, and offer a complement of GCP services.

39. New data center region: Montreal - This new GCP region delivers lower latency for customers in Canada and adjacent geographic areas. Like other Google Cloud regions, it will feature a minimum of three zones, benefit from Google’s global, private fibre network, and offer a complement of GCP services.

40. New data center region: Netherlands - This new GCP region delivers lower latency for customers in Western Europe and adjacent geographic areas. Like other Google Cloud regions, it will feature a minimum of three zones, benefit from Google’s global, private fibre network, and offer a complement of GCP services.

41. Google Container Engine - Managed Nodes - Google Container Engine (GKE) has added Automated Monitoring and Repair of your GKE nodes, letting you focus on your applications while Google ensures your cluster is available and up-to-date.

42. 64 Core machines + more memory - We have doubled the number of vCPUs you can run in an instance from 32 to 64 and up to 416GB of memory per instance.

43. Internal Load balancing (GA) - Internal Load Balancing, now GA, lets you run and scale your services behind a private load balancing IP address which is accessible only to your internal instances, not the internet.

44. Cross-Project Networking (Beta) - Cross-Project Networking (XPN), now in beta, is a virtual network that provides a common network across several Google Cloud Platform projects, enabling simple multi-tenant deployments.

G Suite - Enterprise Collaboration & Productivity

In the past year, we’ve launched 300+ features and updates for G Suite and this week we announced our next generation of collaboration and communication tools.

45. Team Drives (GA for G Suite Business, Education and Enterprise customers) - Team Drives help teams simply and securely manage permissions, ownership and file access for an organization within Google Drive.

46. Drive File Stream (EAP) - Drive File Stream is a way to quickly stream files directly from the cloud to your computer With Drive File Steam, company data can be accessed directly from your laptop, even if you don’t have much space on your hard drive.

47. Google Vault for Drive (GA for G Suite Business, Education and Enterprise customers) - Google Vault for Drive now gives admins the governance controls they need to manage and secure all of their files, including employee Drives and Team Drives. Google Vault for Drive also lets admins set retention policies that automatically keep what’s needed and delete what’s not.

48. Quick Access in Team Drives (GA) - powered by Google’s machine intelligence, Quick Access helps to surface the right information for employees at the right time within Google Drive. Quick Access now works with Team Drives on iOS and Android devices, and is coming soon to the web.

49. Hangouts Meet (GA to existing customers) - Hangouts Meet is a new video meeting experience built on the Hangouts that can run 30-person video conferences without accounts, plugins or downloads. For G Suite Enterprise customers, each call comes with a dedicated dial-in phone number so that team members on the road can join meetings without wifi or data issues.

50. Hangouts Chat (EAP) - Hangouts Chat is an intelligent communication app in Hangouts with dedicated, virtual rooms that connect cross-functional enterprise teams. Hangouts Chat integrates with G Suite apps like Drive and Docs, as well as photos, videos and other third-party enterprise apps.

51. @meet - @meet is an intelligent bot built on top of the Hangouts platform that uses natural language processing and machine learning to automatically schedule meetings for your team with Hangouts Meet and Google Calendar.

52. Gmail Add-ons for G Suite (Developer Preview) - Gmail Add-ons provide a way to surface the functionality of your app or service directly in Gmail. With Add-ons, developers only build their integration once, and it runs natively in Gmail on web, Android and iOS.

53. Edit Opportunities in Google Sheets - with Edit Opportunities in Google Sheets, sales reps can sync a Salesforce Opportunity List View to Sheets to bulk edit data and changes are synced automatically to Salesforce, no upload required.

54. Jamboard - Our whiteboard in the cloud goes GA in May! Jamboard merges the worlds of physical and digital creativity. It’s real time collaboration on a brilliant scale, whether your team is together in the conference room or spread all over the world.

Android & Chrome Devices

Building on the momentum from a growing number of businesses using Chrome digital signage and kiosks, we added new management tools and APIs in addition to introducing support for Android Kiosk apps on supported Chrome devices. 

55. Android Kiosk Apps for Chrome - Android Kiosk for Chrome lets users manage and deploy Chrome digital signage and kiosks for both web and Android apps. And with Public Session Kiosks, IT admins can now add a number of Chrome packaged apps alongside hosted apps.

56. Chrome Kiosk Management Free trial - This free trial gives customers an easy way to test out Chrome for signage and kiosk deployments.

57. Chrome Device Management (CDM) APIs for Kiosks - These APIs offer programmatic access to various Kiosk policies. IT admins can schedule a device reboot through the new APIs and integrate that functionality directly in a third- party console.

58. Chrome Stability API - This new API allows Kiosk app developers to improve the reliability of the application and the system.

Google Cloud Customers

Attendees at Google Cloud Next ‘17 heard stories from many of our valued customers:

59. Colgate - Colgate-Palmolive partnered with Google Cloud and SAP to bring thousands of employees together through G Suite collaboration and productivity tools. The company deployed G Suite to 28,000 employees in less than six months.

60. Disney Consumer Products & Interactive (DCPI) - DCPI is on target to migrate out of its legacy infrastructure this year, and is leveraging machine learning to power next generation guest experiences.

61. eBay - eBay uses Google Cloud technologies including Google Container Engine, Machine Learning and AI for its ShopBot, a personal shopping bot on Facebook Messenger.

62. HSBC - HSBC is one of the world's largest financial and banking institutions and making a large investment in transforming its global IT. The company is working closely with Google to deploy Cloud DataFlow, BigQuery and other data services to power critical proof of concept projects.

63. LUSH - LUSH migrated its global e-commerce site from AWS to GCP in less than six weeks, significantly improving the reliability and stability of its site. LUSH benefits from GCP’s ability to scale as transaction volume surges, which is critical for a retail business. In addition, Google's commitment to renewable energy sources aligns with LUSH's ethical principles.

64. Oden Technologies - Oden was part of Google Cloud’s startup program, and switched its entire platform to GCP from AWS. GCP offers Oden the ability to reliably scale while keeping costs low, perform under heavy loads and consistently delivers sophisticated features including machine learning and data analytics.

65. Planet - Planet migrated to GCP in February, looking to accelerate their workloads and leverage Google Cloud for several key advantages: price stability and predictability, custom instances, first-class Kubernetes support, and Machine Learning technology. Planet also announced the beta release of their Explorer platform.

66. Schlumberger - Schlumberger is making a critical investment in the cloud, turning to GCP to enable high-performance computing, remote visualization and development velocity. GCP is helping Schlumberger deliver innovative products and services to its customers by using HPC to scale data processing, workflow and advanced algorithms.

67. The Home Depot - The Home Depot collaborated with GCP’s Customer Reliability Engineering team to migrate HomeDepot.com to the cloud in time for Black Friday and Cyber Monday. Moving to GCP has allowed the company to better manage huge traffic spikes at peak shopping times throughout the year.

68. Verizon - Verizon is deploying G Suite to more than 150,000 of its employees, allowing for collaboration and flexibility in the workplace while maintaining security and compliance standards. Verizon and Google Cloud have been working together for more than a year to bring simple and secure productivity solutions to Verizon’s workforce.

Google Cloud Partners

We brought together Google Cloud partners from our growing ecosystem across G Suite, GCP, Maps, Devices and Education. Our partnering philosophy is driven by a set of principles that emphasize openness, innovation, fairness, transparency and shared success in the cloud market. Here are some of our partners who were out in force at the show:

69. Accenture - Accenture announced that it has designed a mobility solution for Rentokil, a global pest control company, built in collaboration with Google as part of the partnership announced at Horizon in September.

70. Alooma - Alooma announced the integration of the Alooma service with Google Cloud SQL and BigQuery.

71. Authorized Training Partner Program - To help companies scale their training offerings more quickly, and to enable Google to add other training partners to the ecosystem, we are introducing a new track within our partner program to support their unique offerings and needs.

72. Check Point - Check Point® Software Technologies announced Check Point vSEC for Google Cloud Platform, delivering advanced security integrated with GCP as well as their joining of the Google Cloud Technology Partner Program.

73. CloudEndure - We’re collaborating with CloudEndure to offer a no cost, self-service migration tool for Google Cloud Platform (GCP) customers.

74. Coursera - Coursera announced that it is collaborating with Google Cloud Platform to provide an extensive range of Google Cloud training course. To celebrate this announcement  Coursera is offering all NEXT attendees a 100% discount for the GCP fundamentals class.

75. DocuSign - DocuSign announced deeper integrations with Google Docs.

76. Egnyte - Egnyte announced an enhanced integration with Google Docs that will allow our joint customers to create, edit, and store Google Docs, Sheets and Slides files right from within the Egnyte Connect.

77. Google Cloud Global Partner Awards - We recognized 12 Google Cloud partners that demonstrated strong customer success and solution innovation over the past year: Accenture, Pivotal, LumApps, Slack, Looker, Palo Alto Networks, Virtru, SoftBank, DoIT, Snowdrop Solutions, CDW Corporation, and SYNNEX Corporation.

78. iCharts - iCharts announced additional support for several GCP databases, free pivot tables for current Google BigQuery users, and a new product dubbed “iCharts for SaaS.”

79. Intel - In addition to the progress with Skylake, Intel and Google Cloud launched several technology initiatives and market education efforts covering IoT, Kubernetes and TensorFlow, including optimizations, a developer program and tool kits.

80. Intuit - Intuit announced Gmail Add-Ons, which are designed to integrate custom workflows into Gmail based on the context of a given email.

81. Liftigniter - Liftigniter is a member of Google Cloud’s startup program and focused on machine learning personalization using predictive analytics to improve CTR on web and in-app.

82. Looker - Looker launched a suite of Looker Blocks, compatible with Google BigQuery Data Transfer Service, designed to give marketers the tools to enhance analysis of their critical data.

83. Low interest loans for partners - To help Premier Partners grow their teams, Google announced that capital investment are available to qualified partners in the form of low interest loans.

84. MicroStrategy - MicroStrategy announced an integration with Google Cloud SQL for PostgreSQL and Google Cloud SQL for MySQL.

85. New incentives to accelerate partner growth - We are increasing our investments in multiple existing and new incentive programs; including, low interest loans to help Premier Partners grow their teams, increasing co-funding to accelerate deals, and expanding our rebate programs.

86. Orbitera Test Drives for GCP Partners - Test Drives allow customers to try partners’ software and generate high quality leads that can be passed directly to the partners’ sales teams. Google is offering Premier Cloud Partners one year of free Test Drives on Orbitera.

87. Partner specializations - Partners demonstrating strong customer success and technical proficiency in certain solution areas will now qualify to apply for a specialization. We’re launching specializations in application development, data analytics, machine learning and infrastructure.

88. Pivotal - GCP announced Pivotal as our first CRE technology partner. CRE technology partners will work hand-in-hand with Google to thoroughly review their solutions and implement changes to address identified risks to reliability.

89. ProsperWorks - ProsperWorks announced Gmail Add-Ons, which are designed to integrate custom workflows into Gmail based on the context of a given email.

90. Qwiklabs - This recent acquisition will provide Authorized Training Partners the ability to offer hands-on labs and comprehensive courses developed by Google experts to our customers.

91. Rackspace - Rackspace announced a strategic relationship with Google Cloud to become its first managed services support partner for GCP, with plans to collaborate on a new managed services offering for GCP customers set to launch later this year.

92. Rocket.Chat - Rocket.Chat, a member of Google Cloud’s startup program, is adding a number of new product integrations with GCP including Autotranslate via Translate API, integration with Vision API to screen for inappropriate content, integration to NLP API to perform sentiment analysis on public channels, integration with GSuite for authentication and a full move of back-end storage to Google Cloud Storage.

93. Salesforce - Salesforce announced Gmail Add-Ons, which are designed to integrate custom workflows into Gmail based on the context of a given email.

94. SAP - This strategic partnership includes certification of SAP HANA on GCP, new G Suite integrations and future collaboration on building machine learning features into intelligent applications like conversational apps that guide users through complex workflows and transactions.

95. Smyte - Smyte participated in the Google Cloud startup program and protects millions of actions a day on websites and mobile applications. Smyte recently moved from self-hosted Kubernetes to Google Container Engine (GKE).

96. Veritas - Veritas expanded its partnership with Google Cloud to provide joint customers with 360 Data Management capabilities. The partnership will help reduce data storage costs, increase compliance and eDiscovery readiness and accelerate the customer’s journey to Google Cloud Platform.

97. VMware Airwatch - Airwatch provides enterprise mobility management solutions for Android and continues to drive the Google Device ecosystem to enterprise customers.

98. Windows Partner Program- We’re working with top systems integrators in the Windows community to help GCP customers take full advantage of Windows and .NET apps and services on our platform.

99. Xplenty - Xplenty announced the addition of two new services from Google Cloud into their available integrations: Google Cloud Spanner and Google Cloud SQL for PostgreSQL.

100. Zoomdata - Zoomdata announced support for Google’s Cloud Spanner and PostgreSQL on GCP, as well as enhancements to the existing Zoomdata Smart Connector for Google BigQuery. With these new capabilities Zoomdata offers deeply integrated and optimized support for Google Cloud Platform’s Cloud Spanner, PostgreSQL, Google BigQuery, and Cloud DataProc services.

We’re thrilled to have so many new products and partners that can help all of our customers grow. And as our final announcement for Google Cloud Next ’17 — please save the date for Next 2018: June 4–6 in San Francisco.

I guess that makes it 101. :-)

Read the full article here by Gmail Blog

Announcing TensorFlow 1.0

Originally posted on the Google Developer Blog

In just its first year, TensorFlow has helped researchers, engineers, artists, students, and many others make progress with everything from language translation to early detection of skin cancer and preventing blindness in diabetics. We're excited to see people using TensorFlow in over 6000 open source repositories online.

Today, as part of the first annual TensorFlow Developer Summit, hosted in Mountain View and livestreamed around the world, we're announcing TensorFlow 1.0:

It's faster: TensorFlow 1.0 is incredibly fast! XLA lays the groundwork for even more performance improvements in the future, and tensorflow.org now includes tips & tricksfor tuning your models to achieve maximum speed. We'll soon publish updated implementations of several popular models to show how to take full advantage of TensorFlow 1.0 - including a 7.3x speedup on 8 GPUs for Inception v3 and 58x speedup for distributed Inception v3 training on 64 GPUs!

It's more flexible: TensorFlow 1.0 introduces a high-level API for TensorFlow, with tf.layers, tf.metrics, and tf.losses modules. We've also announced the inclusion of a new tf.keras module that provides full compatibility with Keras, another popular high-level neural networks library.

It's more production-ready than ever: TensorFlow 1.0 promises Python API stability (details here), making it easier to pick up new features without worrying about breaking your existing code.

Other highlights from TensorFlow 1.0:

• Python APIs have been changed to resemble NumPy more closely. For this and other backwards-incompatible changes made to support API stability going forward, please use our handy migration guide and conversion script.
• Experimental APIs for Javaand Go
• Higher-level API modules tf.layers, tf.metrics, and tf.losses - brought over from tf.contrib.learnafter incorporating skflowand TF Slim
• Experimental release of XLA, a domain-specific compiler for TensorFlow graphs, that targets CPUs and GPUs. XLA is rapidly evolving - expect to see more progress in upcoming releases.
• Introduction of the TensorFlow Debugger (tfdbg), a command-line interface and API for debugging live TensorFlow programs.
• New Android demos for object detection and localization, and camera-based image stylization.
• Installation improvements: Python 3 docker images have been added, and TensorFlow's pip packages are now PyPI compliant. This means TensorFlow can now be installed with a simple invocation of pip install tensorflow.

We're thrilled to see the pace of development in the TensorFlow community around the world. To hear more about TensorFlow 1.0 and how it's being used, you can watch the TensorFlow Developer Summit talks on YouTube, covering recent updates from higher-level APIs to TensorFlow on mobile to our new XLA compiler, as well as the exciting ways that TensorFlow is being used:

Click herefor a link to the livestream and video playlist (individual talks will be posted online later in the day).

The TensorFlow ecosystem continues to grow with new techniques like Foldfor dynamic batching and tools like the Embedding Projector along with updatesto our existing tools like TensorFlow Serving. We're incredibly grateful to the community of contributors, educators, and researchers who have made advances in deep learning available to everyone. We look forward to working with you on forums like GitHub issues, Stack Overflow, @TensorFlow, the discuss@tensorflow.orggroup, and at future events.

By Amy McDonald Sandjideh, Technical Program Manager, TensorFlow

Read the full article here by Google Open Source Blog

Introducing Associate Android Developer Certification by Google

Posted by JP Souchak, Program Manager

The Associate Android Developer Certification program was announced at Google I/O 2016, and launched a few months later. Since then, over 322 Android developers spanning 61 countries have proven their competency and earned the title of Google Certified Associate Android Developer.

To establish a professional standard for what it means to be an Associate Android developer in the current job market, Google created this certification, which allows us to recognize developers who have proven themselves to uphold that standard.

We conducted a job task analysis to determine the required competencies and content of the certification exam. Through field research and interviews with experts, we identified the knowledge, work practices, and essential skills expected of an Associate Android developer.

The certification process consists of a performance-based exam and an exit interview. The certification fee includes three exam attempts. The cost for certification is $149 USD, or 6500 INR if you reside in India. After payment, the exam will be available for download, and you have 48 hours to complete and submit it for grading.

In the exam, you will implement missing features and debug an Android app using Android Studio. If you pass, you will undergo an exit interview where, you will answer questions about your exam and demonstrate your knowledge of Associate Android Developer competencies.

Check out this short video for a quick overview of the Associate Android Developer certification process:

Earning your AAD Certification signifies that you possess the skills expected of an Associate Android developer, as determined by Google. You can showcase your credential on your resume and display your digital badge on your social media profiles. As a member of the AAD Alumni Community, you will also have access to program benefits focused on increasing your visibility as a certified developer.

Test your Android development skills and earn the title of Google Certified Associate Android Developer. Visit http://ift.tt/2krNVG3 to get started!

Read the full article here by Android Developers Blog

Start building Actions on Google

Posted by Jason Douglas, PM Director for Actions on Google

The Google Assistant brings together all of the technology and smarts we've been building for years, from the Knowledge Graph to Natural Language Processing. To be a truly successful Assistant, it should be able to connect users across the apps and services in their lives. This makes enabling an ecosystem where developers can bring diverse and unique services to users through the Google Assistant really important.

In October, we previewed Actions on Google, the developer platform for the Google Assistant. Actions on Google further enhances the Assistant user experience by enabling you to bring your services to the Assistant. Starting today, you can build Conversation Actions for Google Home and request to become an early access partner for upcoming platform features.

Conversation Actions for Google Home

Conversation Actions let you engage your users to deliver information, services, and assistance. And the best part? It really is a conversation -- users won't need to enable a skill or install an app, they can just ask to talk to your action. For now, we've provided two developer samples of what's possible, just say "Ok Google, talk to Number Genie " or try "Ok Google, talk to Eliza' for the classic 1960s AI exercise.

You can get started today by visiting the Actions on Google website for developers. To help create a smooth, straightforward development experience, we worked with a number of development partners, including conversational interaction development tools API.AI and Gupshup, analytics tools DashBot and VoiceLabs and consulting companies such as Assist, Notify.IO, Witlingo and Spoken Layer. We also created a collection of samples and voice user interface (VUI) resources or you can check out the integrations from our early access partners as they roll out over the coming weeks.

Introduction to Conversation Actions by Wayne Piekarski

Coming soon: Actions for Pixel and Allo + Support for Purchases and Bookings

Today is just the start, and we're excited to see what you build for the Google Assistant. We'll continue to add more platform capabilities over time, including the ability to make your integrations available across the various Assistant surfaces like Pixel phones and Google Allo. We'll also enable support for purchases and bookings as well as deeper Assistant integrations across verticals. Developers who are interested in creating actions using these upcoming features should register for our early access partner program and help shape the future of the platform.

Build, explore and let us know what you think about Actions on Google! And to say in the loop, be sure to sign up for our newsletter, join our Google+ community, and use the “actions-on-google” tag on StackOverflow.

Read the full article here by Android Developers Blog

Open-sourcing DeepMind Lab

Originally posted on DeepMind Blog

DeepMind's scientific mission is to push the boundaries of AI, developing systems that can learn to solve any complex problem without needing to be taught how. To achieve this, we work from the premise that AI needs to be general. Agents should operate across a wide range of tasks and be able to automatically adapt to changing circumstances. That is, they should not be pre-programmed, but rather, able to learn automatically from their raw inputs and reward signals from the environment. There are two parts to this research program: (1)  designing ever-more intelligent agents capable of more-and-more sophisticated cognitive skills, and (2) building increasingly complex environments where agents can be trained and evaluated.

The development of innovative agents goes hand in hand with the careful design and implementation of rationally selected, flexible and well-maintained environments. To that end, we at DeepMind have invested considerable effort toward building rich simulated environments to serve as  “laboratories” for AI research. Now we are open-sourcing our flagship platform,  DeepMind Lab, so the broader research community can make use of it.

DeepMind Lab is a fully 3D game-like platform tailored for agent-based AI research. It is observed from a first-person viewpoint, through the eyes of the simulated agent. Scenes are rendered with rich science fiction-style visuals. The available actions allow agents to look around and move in 3D. The agent’s “body” is a floating orb. It levitates and moves by activating thrusters opposite its desired direction of movement, and it has a camera that moves around the main sphere as a ball-in-socket joint tracking the rotational look actions. Example tasks include collecting fruit, navigating in mazes, traversing dangerous passages while avoiding falling off cliffs, bouncing through space using launch pads to move between platforms, playing laser tag, and quickly learning and remembering random procedurally generated environments. An illustration of how agents in DeepMind Lab perceive and interact with the world can be seen below:

At each moment in time, agents observe the world as an image, in pixels, rendered from their own first-person perspective. They also may receive a reward (or punishment!) signal. The agent can activate its thrusters to move in 3D and can also rotate its viewpoint along both horizontal and vertical axes.

Artificial general intelligence research in DeepMind Lab emphasizes navigation, memory, 3D vision from a first person viewpoint, motor control, planning, strategy, time, and fully autonomous agents that must learn for themselves what tasks to perform by exploring their environment. All these factors make learning difficult. Each are considered frontier research questions in their own right. Putting them all together in one platform, as we have, represents a significant new challenge for the field.


DeepMind Lab is highly customisable and extendable. New levels can be authored with off-the-shelf editor tools. In addition, DeepMind Lab includes an interface for programmatic level-creation. Levels can be customised with gameplay logic, item pickups, custom observations, level restarts, reward schemes, in-game messages and more. The interface can be used to create levels in which novel map layouts are generated on the fly while an agent trains. These features are useful in, for example, testing how an agent copes with unfamiliar environments. Users will be able to add custom levels to the platform via GitHub. The assets will be hosted on GitHub alongside all the code, maps and level scripts. Our hope is that the community will help us shape and develop the platform going forward.



DeepMind Lab has been used internally at DeepMind for some time (example). We believe it has already had a significant impact on our thinking concerning numerous aspects of intelligence, both natural and artificial. However, our efforts so far have only barely scratched the surface of what is possible in DeepMind Lab. There are opportunities for significant contributions still to be made in a number of mostly still untouched research domains now available through DeepMind Lab, such as navigation, memory and exploration.

As well as facilitating agent evaluation, there are compelling reasons to think that it may be fundamentally easier to develop intelligence in a 3D world, observed from a first-person viewpoint, like DeepMind Lab. After all, the only known examples of general-purpose intelligence in the natural world arose from a combination of evolution, development, and learning, grounded in physics and the sensory apparatus of animals. It is possible that a large fraction of animal and human intelligence is a direct consequence of the richness of our environment, and unlikely to arise without it. Consider the alternative: if you or I had grown up in a world that looked like Space Invaders or Pac-Man, it doesn’t seem likely we would have achieved much general intelligence!

Read the full paper here.

Access DeepMind's GitHub repository here.

By Charlie Beattie, Joel Leibo, Stig Petersen and Shane Legg, DeepMind Team

Read the full article here by Google Open Source Blog

Docker + Dataflow = happier workflows

When I first saw the Google Cloud Dataflow monitoring UI -- with its visual flow execution graph that updates as your job runs, and convenient links to the log messages -- the idea came to me. What if I could take that UI, and use it for something it was never built for? Could it be connected with open source projects aimed at promoting reproducible scientific analysis, like Common Workflow Language (CWL) or Workflow Definition Language (WDL)?

Screenshot of a Dockerflow workflow for DNA sequence analysis.

In scientific computing, it’s really common to submit jobs to a local high-performance computing (HPC) cluster. There are tools to do that in the cloud, like Elasticluster and Starcluster. They replicate the local way of doing things, which means they require a bunch of infrastructure setup and management that the university IT department would otherwise do. Even after you’re set up, you still have to ssh into the cluster to do anything. And then there are a million different choices for workflow managers, each unsatisfactory in its own special way.

By day, I’m a product manager. I hadn’t done any serious coding in a few years. But I figured it shouldn’t be that hard to create a proof-of-concept, just to show that the Apache Beam API that Dataflow implements can be used for running scientific workflows. Now, Dataflow was created for a different purpose, namely, to support scalable data-parallel processing, like transforming giant data sets, or computing summary statistics, or indexing web pages. To use Dataflow for scientific workflows would require wrapping up shell steps that launch VMs, run some code, and shuttle data back and forth from an object store. It should be easy, right?

It wasn’t so bad. Over the weekend, I downloaded the Dataflow SDK, ran the wordcount examples, and started modifying. I had a “Hello, world” proof-of-concept in a day.

To really run scientific workflows would require more, of course. Varying VM shapes, a way to pass parameters from one step to the next, graph definition, scattering and gathering, retries. So I shifted into prototyping mode.

I created a new GitHub project called Dockerflow. With Dockerflow, workflows can be defined in YAML files. They can also be written in pretty compact Java code. You can run a batch of workflows at once by providing a CSV file with one row per workflow to define the parameters.

Dataflow and Docker complement each other nicely:

• Dataflow provides a fully managed service with a nice monitoring interface, retries,  graph optimization and other niceties.
• Docker provides portability of the tools themselves, and there's a large library of packaged tools already available as Docker images.

While Dockerflow supports a simple YAML workflow definition, a similar approach could be taken to implement a runner for one of the open standards like CWL or WDL.

To get a sense of working with Dockerflow, here’s “Hello, World” written in YAML:

defn:
  name: HelloWorkflow
steps:
- defn:
    name: Hello
    inputParameters:
      name: message
      defaultValue: Hello, World!
    docker:
      imageName: ubuntu
      cmd: echo $message

And here’s the same example written in Java:

public class HelloWorkflow implements WorkflowDefn {
  @Override
  public Workflow createWorkflow(String[] args) throws IOException {
    Task hello =
        TaskBuilder.named("Hello").input("message", “Hello, World!”).docker(“ubuntu”).script("echo $message").build();
    return TaskBuilder.named("HelloWorkflow").steps(hello).args(args).build();
  }
}

Dockerflow is just a prototype at this stage, though it can run real workflows and includes many nice features, like dry runs, resuming failed runs from mid-workflow, and, of course, the nice UI. It uses Cloud Dataflow in a way that was never intended -- to run scientific batch workflows rather than large-scale data-parallel workloads. I wish I’d written it in Python rather than Java. The Dataflow Python SDK wasn’t quite as mature when I started.

Which is all to say, it’s been a great 20% project, and the future really depends on whether it solves a problem people have, and if others are interested in improving on it. We welcome your contributions and comments! How do you run and monitor scientific workflows today?

By Jonathan Bingham, Google Genomics and Verily Life Sciences

Read the full article here by Google Open Source Blog

Google Open Source Report Card

Open source software enables Google to build things quickly and efficiently without reinventing the wheel, allowing us to focus on solving new problems. We stand on the shoulders of giants and we know it. This is why we support open source and make it easy for Googlers to release the projects they’re working on internally as open source.

Today we’re sharing our first Open Source Report Card, highlighting our most popular projects, sharing a few statistics and detailing some of the projects we’ve released in 2016.

We’ve open sourced over 20 million lines of code to date and you can find a listing of some of our best known project releases on our website. Here are some of our most popular projects:

• Android - a software stack for mobile devices that includes an operating system, middleware and key applications.
• Chromium - a project encompassing Chromium, the software behind Google Chrome, and Chromium OS, the software behind Google Chrome OS devices.
• Angular - a web application framework for JavaScript and Dart focused on developer productivity, speed and testability.
• TensorFlow - a library for numerical computation using data flow graphics with support for scalable machine learning across platforms from data centers to embedded devices.
• Go - a statically typed and compiled programming language that is expressive, concise, clean and efficient.
• Kubernetes - a system for automating deployment, operations and scaling of containerized applications.
• Polymer - a lightweight library built on top of Web Components APIs for building encapsulated re-usable elements in web applications.
• Protobuf - an extensible, language-neutral and platform-neutral mechanism for serializing structured data.
• Guava - a set of Java core libraries that includes new collection types (such as multimap and multiset), immutable collections, a graph library, functional types, an in-memory cache, and APIs/utilities for concurrency, I/O, hashing, primitives, reflection, string processing and much more.
• Yeoman - a robust and opinionated set of scaffolding tools including libraries and a workflow that can help developers quickly build beautiful and compelling web applications.

While it’s difficult to measure the full scope of open source at Google, we can use the subset of projects that are on GitHub to gather some interesting data. Today our GitHub footprint includes over 84 organizations and 3,499 repositories, 773 of which were created this year.

Googlers use countless languages from Assembly to XSLT, but what are their favorites? GitHub flags the most heavily used language in a repository and we can use that to find out. A survey of GitHub repositories shows us these are some of the languages Googlers use most often:

• JavaScript
• Java
• C/C++
• Go
• Python
• TypeScript
• Dart
• PHP
• Objective-C
• C#

Many things can be gleaned using the open source GitHub dataset on BigQuery, like usage of tabs versus spaces and the most popular Go packages. What about how many times Googlers have committed to open source projects on GitHub? We can search for Google.com email addresses to get a baseline number of Googler commits. Here’s our query:

SELECT count(*) as n
FROM [bigquery-public-data:github_repos.commits]
WHERE committer.date > '2016-01-01 00:00'
AND REGEXP_EXTRACT(author.email, r'.*@(.*)') = 'google.com'

With this, we learn that Googlers have made 142,527 commits to open source projects on GitHub since the start of the year. This dataset goes back to 2011 and we can tweak this query to find out that Googlers have made 719,012 commits since then. Again, this is just a baseline number as it doesn’t count commits made with other email addresses.

Looking back at the projects we’ve open-sourced in 2016 there’s a lot to be excited about. We have released open source software, hardware and datasets. Let’s take a look at some of this year’s releases.

Seesaw
Seesaw is a Linux Virtual Server (LVS) based load balancing platform developed in Go by our Site Reliability Engineers. Seesaw, like many projects, was built to scratch our own itch.

From our blog post announcing its release: “We needed the ability to handle traffic for unicast and anycast VIPs, perform load balancing with NAT and DSR (also known as DR), and perform adequate health checks against the backends. Above all we wanted a platform that allowed for ease of management, including automated deployment of configuration changes.”

Vendor Security Assessment Questionnaire (VSAQ)
We assess the security of hundreds of vendors every year and have developed a process to automate much of the initial information gathering with VSAQ. Many vendors found our questionnaires intuitive and flexible, so we decided to shared them. The VSAQ Framework includes four extensible questionnaire templates covering web applications, privacy programs, infrastructure as well as physical and data center security. You can learn more about it in our announcement blog post.

OpenThread
OpenThread, released by Nest, is a complete implementation of the Thread protocol for connected devices in the home. This is especially important because of the fragmentation we’re seeing in this space. Development of OpenThread is supported by ARM, Microsoft, Qualcomm, Texas Instruments and other major vendors.

Magenta
Can we use machine learning to create compelling art and music? That’s the question that animates Magenta, a project from the Google Brain team based on TensorFlow. The aim is to advance the state of the art in machine intelligence for music and art generation and build a collaborative community of artists, coders and machine learning researchers. Read the release announcement for more information.

Omnitone
Virtual reality (VR) isn’t nearly as immersive without spatial audio and much of VR development is taking place on proprietary platforms. Omnitone is an open library built by members of the Chrome Team that brings spatial audio to the browser. Omnitone builds on standard Web Audio APIs to deliver an immersive experience and can be used alongside projects like WebVR. Find out more in our blog post announcing the project’s release.

Science Journal
Today’s smartphones are packed with sensors that can tell us interesting things about the world around us. We launched Science Journal to help educators, students and citizen scientists tap into those sensors. You can learn more about the project in our announcement blog post.

Cartographer
Cartographer is a library for real-time simultaneous localization and mapping (SLAM) in 2D and 3D with Robot Operating System (ROS) support. Combining data from a variety of sensors, this library computes positioning and maps surroundings. This is a key element of self-driving cars, UAVs and robotics as well as efforts to map the insides of famous buildings. More information on Cartographer can be found in our blog post announcing its release.

This is just a small sampling of what we’ve released this year. Follow the Google Open Source Blog to stay apprised of Google’s open source software, hardware and data releases.

By Josh Simmons, Open Source Programs Office

Read the full article here by Google Open Source Blog