Google details an AI model called MetNet, better than NOAA for predicting precipitation

Venturebeat | March 26, 2020

  • Google researchers published a paper that details an AI system MetNet that can predict precipitation up to eight hours.

  • According to researchers it outperforms the current state-of-the-art physics model in use by the U.S. National Oceanic and Atmospheric Administration (NOAA) and that it makes a prediction over the entire U.S. in seconds as opposed to an hour.

  • One key advantage of MetNet is that it’s optimized for dense and parallel computation and well-suited for running on specialty hardware such as Google-designed tensor processing units (TPUs).


In a blog post and accompanying paper, researchers at Google detail an AI system — MetNet — that can predict precipitation up to eight hours into the future. They say that it outperforms the current state-of-the-art physics model in use by the U.S. National Oceanic and Atmospheric Administration (NOAA) and that it makes a prediction over the entire U.S. in seconds as opposed to an hour.
 

It builds on previous work from Google, which created an AI system that ingested satellite images to produce forecasts with a roughly one-kilometer resolution and a latency of only 5-10 minutes. And while it’s early days, it could lay the runway for a forecasting tool that could help businesses, residents, and local governments better prepare for inclement weather.
 

Learn more: Google designs open source framework that reduces AI model training cost by 80%
 

MetNet takes a data-driven and physics-free approach to weather modeling, meaning it learns to approximate atmospheric physics from examples and not by incorporating prior knowledge. Specifically, it uses precipitation estimates derived from ground-based radar stations and measurements from NOAA’s Geostationary Operational Environmental Satellite that provide a top-down view of clouds in the atmosphere. Both sources cover the continental U.S., providing image-like inputs that can be processed by the model.
 

MetNet is executed for every 64-by-64-kilometer square covering the U.S. at a 1-kilometer resolution. As the paper’s authors explain, the physical coverage corresponding to each output region is much larger — a 1,024-by-1,024-kilometer square — since the model must take into account the possible motion of the clouds and precipitation fields over time. For example, to make a prediction 8 hours ahead, assuming that clouds move up to 60 kilometers per hour, MetNet needs 480 kilometers (60 x 8) of context.
 

MetNet’s spatial downsampler component decreases the memory consumption while finding and retaining the relevant weather patterns, and its temporal encoder encodes snapshots from the previous 90 minutes of input data in 15-minute segments. The output is a discrete probability distribution estimating the probability of a given rate of precipitation for each square kilometer in the continental U.S.

One key advantage of MetNet is that it’s optimized for dense and parallel computation and well-suited for running on specialty hardware such as Google-designed tensor processing units (TPUs). This allows predictions to be made in parallel in a matter of seconds, whether for a specific location like New York City or for the entire U.S.


The researchers tested MetNet on a precipitation rate forecasting benchmark and compared the results with two baselines — the NOAA High Resolution Rapid Refresh (HRRR) system, which is the physical weather forecasting model currently operational in the U.S., and a baseline model that estimates the motion of the precipitation field, or optical field. They report that in terms of F1-score at a precipitation rate threshold of 1 millimeter per hour, which corresponds to light rain, MetNet outperformed both the flow-based model and HRRR system for timescales up to 8 hours ahead. Recently Google AI team explains how its audio recorder app leverages on-device ML.


We are actively researching how to improve global weather forecasting, especially in regions where the impacts of rapid climate change are most profound.

-Nal Kalchbrenner and Casper Sønderby, Google research scientists


Further they said, “While we demonstrate the present MetNet model for the continental U.S., it could be extended to cover any region for which adequate radar and optical satellite data are available. The work presented here is a small stepping stone in this effort that we hope leads to even greater improvements through future collaboration with the meteorological community.”

 

Learn more: Google launches TensorFlow constrained optimization

Spotlight

AI has had a long and rich history with games of all sorts. From early days of AI systems learning how to play chess and checkers to modern day AI defeating some of the best players in a range of different games such as Go or Doda. And gaming  is a huge industry accounting for annual revenue for digital games of $25.43 Billion in the US alone. In this infographic Cognilytica explores various ways AI is being used in gaming.


Other News
AI TECH

Pega Delivers New Intelligent, Low-Code Capabilities for AI-Powered Decisioning and Workflow Automation

Pegasystems | March 23, 2022

Pegasystems Inc. , the software company that crushes business complexity, today announced the latest edition of its Pega Infinity™ software suite that helps empower organizations to quickly deploy apps, create smarter workflows, and deliver better experiences for businesses and their customers. With low-code innovation, applied AI and automation, and a modern cloud architecture, Pega Infinity version 8.7 helps organizations innovate faster at scale while creating better total experiences for all end users. An increasingly competitive marketplace made more challenging by the pandemic-induced labor shortage has resulted in a pressing need to retain and empower employees while also meeting and surpassing customers' expectations. Gartner defines this as total experience (TX) – creating optimal environments for employees and customers. To do so, organizations need low-code, intelligent solutions that help employees (from pro developers to business users) innovate quickly and work effectively while giving customers optimal, personalized experiences during every interaction. Available now, Pega Infinity 8.7 helps organizations improve TX with a seamlessly integrated set of capabilities that accelerate digital transformation with the speed, scale, and flexibility needed to overcome any business challenge or need. New enhancements to Pega Platform™, Pega Customer Service™, and Pega Customer Decision Hub™ enable businesses leaders to: Optimize outcomes with AI-powered capabilities in Pega Customer Decision Hub to better engage with customers, including: Customer Profile Viewer: This new feature lets users deep dive into individual customer's propensities, behaviors, suppressions, and potential value. They can also explore customers' omnichannel interaction history to better optimize responses, proactively detect future issues, and simulate next-best-action strategy performance. Action Performance Tracker: Also new in Pega Infinity 8.7, this feature helps users proactively monitor offer performance and determine the success of next-best-action recommendations. Automated alerts inform users of anomalies and help prioritize triage, streamlining the troubleshooting process. Adaptive gradient boosting: These online learning algorithms increase predictive accuracy by allowing users to understand the impact of each model feature on overall performance. This helps ensure success before deploying to production for the most relevant, empathetic customer experiences possible. Provide faster, more personalized customer service with enhancements to Pega Customer Service for digital messaging, pre-built service workflows, and AI-powered tools for agents, such as: Pega Voice AI and Messaging AI: As previewed last month, these Conversational AI capabilities serve as copilots for agents, listening to live phone and chat conversations, analyzing intent, recommending next steps, improving script adherence, and doing data entry for agents, which reduces average handle time and improves agent and customer experiences. New digital messaging capabilities: Asynchronous web messaging lets customers leave a web chat and return to it later without restarting the conversation. A new mobile messaging software development kit (SDK) lets organizations embed web messaging directly into iOS and Android mobile apps, providing more digital engagement options to customers. Industry-specific microjourneys: New pre-built service workflows for the communications, financial services, and healthcare industries help deploy and implement automated service processes faster. Work smarter and more efficiently within Pega Platform with a future-proof architecture that offers applied AI and low-code innovation, such as: Enhanced Pega Process AI: New capabilities added to Pega Process AI boost efficiency by predicting which cases will and will not be successfully fulfilled, as well as learn from historical data to automate work. Users can proactively address customer needs using data and insights from each event as it happens. Updated Robot Studio: With a complete redesign of Robot Studio, developers can reduce time spent on creating bots and ongoing maintenance. Upgrades include developer usability improvements from intelligent recording to easier testing workflows. Improved low-code UX: Developers can deploy rich user experiences (UX) with more speed and control with a modern, single-page application architecture for more responsive user interfaces (UI). Pro developers can orchestrate multiple experiences from the same app definition while citizen developers can quickly create quality UIs with built-in guardrails. Pega Infinity unifies customer engagement, customer service, and intelligent automation capabilities within a single platform to help accelerate digital transformation across entire global enterprises. By engaging customers with the right message, at the right time, across any channel, clients can improve customer satisfaction, increase customer lifetime value, and boost organizational productivity. Most importantly, Pega's low-code environment helps reduce business complexity – resulting in accelerated deployments, faster iteration, and greater customer and employee satisfaction. "The way we do business has changed. Today's leaders know providing great customer experiences is no longer enough. Instead, they need to refocus their efforts to also include great employee experiences. "By implementing solutions that help create optimal total experience, organizations gain a tremendous advantage that not only creates better customer experiences, but also empowers employees to be productive and innovate faster. These new updates to Pega Infinity are helping organizations drive truly transformative outcomes for employees and customers alike." Kerim Akgonul, chief product officer, Pegasystems About Pegasystems Pega delivers innovative software that crushes business complexity so our clients can make better decisions and get work done. We help the world's leading brands solve their biggest business challenges: maximizing customer lifetime value, streamlining customer service, and boosting operational efficiency. Pega technology is powered by real-time AI and intelligent automation, while our scalable architecture and low-code platform help enterprises adapt to rapid change and transform for tomorrow.

Read More

SOFTWARE

PixelPlex Launches IT Consulting Services for Businesses of All Sizes

PixelPlex | March 14, 2022

PixelPlex, a full-cycle software developer, offers IT consulting services to help clients investigate the feasibility of their ideas, select technologies to implement them, and create sophisticated yet easy-to-use solutions. To date, PixelPlex has successfully launched over 450 projects. With such extensive experience of working with companies of various sizes, PixelPlex points out several major pain points that most enterprises face. The biggest problem is low efficiency and security issues associated with using outdated systems. Another one is the lack of knowledge about technologies, which leads to the low performance of implemented solutions. The last common problem is the difficulty of designing a project development strategy. After analyzing these issues, PixelPlex developed and launched IT consulting services. PixelPlex can work on new projects or join existing ones. They start the process by studying the client's business case and analyzing ideas. Then, the consultants model the solution's architecture and infrastructure, provide estimation and project roadmapping, and offer development services. The team also identifies issues and solves technical puzzles to ensure that the final solution is free of vulnerabilities, highly efficient, and performs well. PixelPlex also shared the results of two recent projects — LaneAxis and Proleague.gg. LaneAxis aimed to introduce blockchain into their shipper-to-carrier direct freight network to reduce negotiation times and make negotiations transparent, improve shipment traceability, and speed up problem resolution processes. PixelPlex proposed an effective solution, namely how to capture value with blockchain, replace third-party payment processors, and improve the platform's speed and security. The consultants created a tokenomics model that included the token system, payment methods, stablecoin purchase process, and native token trading. Proleague.gg approached PixelPlex with the idea of developing the Play2Earn ecosystem. The consultants helped the client work out a clear project development strategy and prepared a well-thought-out vision and scope document. The latter outlined main features, the scope for the MVP, tokenomics model, monetization opportunities, and the tech stack. About PixelPlex PixelPlex is a full-cycle consulting and software development company with cross-domain expertise in blockchain, DeFi, FinTech, AI, and AR/VR. The company is a trusted partner for startups, SMEs, and enterprises. PixelPlex's client list includes well-known companies such as BMW, Microsoft, Swisscom, and Kakao.

Read More

Acronis Recognized as a Visionary in Gartner 2020 Magic Quadrant for Data Center Backup and Recovery Solutions

Acronis | July 28, 2020

Acronis, a global leader in cyber protection, was positioned in the Visionaries quadrant in the Gartner 2020 Magic Quadrant for Data Center Backup and Recovery Solutions. According to the report, “the move toward public cloud, heightened concerns over ransomware, and complexities associated with backup and data management are forcing I&O leaders to rearchitect their backup infrastructure and explore alternative solutions. This research provides analyses of backup and recovery vendors.”

Read More

Kira Systems' Response to COVID-19

Kira Systems | March 13, 2020

Kira Systems understands that the coronavirus (COVID-19) is impacting everyone around the world. In the interests of our customers, our staff, their families and their communities, we have taken a number of steps to do our part in reducing the risk associated with this outbreak and ensuring we can continue to service our customers through this challenging time

Read More

Spotlight

AI has had a long and rich history with games of all sorts. From early days of AI systems learning how to play chess and checkers to modern day AI defeating some of the best players in a range of different games such as Go or Doda. And gaming  is a huge industry accounting for annual revenue for digital games of $25.43 Billion in the US alone. In this infographic Cognilytica explores various ways AI is being used in gaming.

Resources