TEPCO Selects Via Science for Strategic Investment

Japan’s largest electric power supplier, Tokyo Electric Power Company Holdings, Inc. (TEPCO), has selected Via Science for its first strategic venture investment in machine learning and analytics.

Cambridge, Mass. – December 6, 2016 – Via Science, Inc. a leader in causal analytics, today announced a minority investment by Japan’s largest electric power supplier, Tokyo Electric Power Company Holdings, Inc. (TEPCO). TEPCO’s strategic investment includes capital, as well as sharing subject matter expertise and data to improve the reliability and operating efficiencies of power suppliers and their customers through advanced software analytics applications.

Already Japan’s largest electric power supplier, TEPCO is investing heavily in new technologies that even further improve the quality and variety of services it provides to its 29 million customers.

“Analytics is critical to improving the operational efficiency of energy companies today and allowing them to stay ahead of regulatory and technological disruptions. TEPCO’s investment is a validation of Via Science’s leadership in causal analytics for applications such as identifying critical infrastructure that is at high risk for power outages, years in advance. We are truly honored to be their first venture investment in machine learning and analytics,” said Colin Gounden, CEO of Via Science.

Through this investment, TEPCO will gain experience in Via Science’s proprietary causal analytics software, which goes beyond traditional “black box” machine learning to provide transparency into the rationale for recommendations to improve maintenance, service staffing, cybersecurity and resiliency planning. TEPCO’s customers will also benefit from Via Science’s expertise, as new technologies enable improved reliability and new value-added services.

“To find the team and technology best able to handle the unique challenges in the energy industry, our innovation group evaluated large and small analytics companies across North America, Europe and Asia. Via Science’s direct experience with US energy agencies and their ability to support our challenges, such as high volumes of data, regulatory transparency requirements and very long-term risk planning needs for multi-billion dollar investments, made them the best match for TEPCO and our customers,” explained Hirokazu Yamaguchi, Executive General Manager, Global Innovation and Investments at TEPCO.

Update: Pierce Atwood congratulates Via Science on securing TEPCO’s strategic investment.

About Via Science

Via Science creates software applications that identify and prioritize risks to critical infrastructure using a combination of high-performance computing, machine learning and causal analytics. Headquartered in Cambridge, Massachusetts, with its research center in Montreal, Quebec, Via Science has developed its risk assessment algorithms in conjunction with some of the world’s leading companies and government agencies in energy and security. The company has been featured in Wired Magazine, The Wall Street Journal and Forbes for its expertise in causal analytics software applications. For more information, please visit www.viascience.com.

About Tokyo Electric Power Company Holdings, Inc.

Tokyo Electric Power Company Holdings, Inc. (TSE: 9501), headquartered in Tokyo, Japan, is the largest utility in Japan serving more than 29 million homes and businesses. Worldwide the company has more than 74 subsidiaries and affiliates in 8 countries and employs approximately 43,330 people. Consolidated revenue for the fiscal year ending March 31, 2016, totaled 6.8 trillion Japanese yen. The company was established in 1951 and is listed on the First Section of the Tokyo Stock Exchange. For more information, visit http://www.tepco.co.jp/en/corpinfo/index-e.html.

The Igarapé Institute, Via Science Launch Public Safety Digital Application

The Igarapé Institute and Via Science formed a partnership to develop CrimeRadar, a digital application accessible on mobile and desktop browsers that uses predictive analytics to forecast the probability of crime throughout Rio de Janeiro.

CrimeRadar uses Via Science’s microservices architecture, Focus™ to generate high-resolution risk ratings for 250-square meter sectors across metropolitan Rio. Focus™ analyzes five years (or 14 million events) of crime and regional data, a user’s current or future location and a timestamp to score the sector. Scores account for seasonality (time of day, day of the week, etc.), sector population and immediately adjacent sectors and time periods. Once a score is generated, the app displays a color-coded visual of potential crime levels within a 250-square meter radius of the specified location.

The Igarapé Institute provided expert knowledge about the region and worked with Rio’s Institute of Public Safety (ISP), considered the most comprehensive public safety dataset available, to gather accurate historical data for the app. Via Science developed the risk-scoring algorithms using advanced machine learning and proprietary software tools.

“The seasonality of crime was a critical component for us to include because it gives a more accurate representation of what the crime level will most likely be depending on the time of day and day of the week. Speed is also a huge factor when creating an app because people expect that when they open an app, it will work fast and exactly as it should without any delay. This can be especially challenging when you have millions of people trying to access the same information at once. We’re very proud of what we have been able to accomplish in creating CrimeRadar with Igarapé,” said Via Science’s CEO, Colin Gounden.

A number of factors play into the timing of CrimeRadar’s release, including an increase in crime rates in Rio de Janeiro, newly accessible crime data, advancements in and availability of analytical tools and this summer’s Olympic Games.

“Preventing and reducing criminal violence requires a change of mindset along with leadership, better data and targeted public policies focused on at-risk places, people and behaviors. At Igarapé, we are regularly designing digital tools to empower citizens to better understand (and engage in) their own safety. With the increase in violence in Brazil especially, we want residents and visitors to have the ability to use intuitive tools to make them feel safer, which is the goal of CrimeRadar,” said Robert Muggah, Research Director at the Igarapé Institute.

Roughly 2,000 people were killed in Rio de Janeiro last year, according to the Igarapé Institute. Crime varies by location, time of day, time of year and myriad other factors. Even the safest neighborhoods experience some level of crime. CrimeRadar aims to provide greater awareness by indicating data-based trends, and does not claim to state exactly when and where crime will occur, which can be impossible to predict. The Igarapé Institute and Via Science both recommend exercising caution when traveling to new and unfamiliar locations.

Via Science Releases Microservices Architecture, Focus™

Via Science is excited to share details regarding the release of its new microservices architecture, Focus™.

Focus™ is a robust architecture that ingests, transforms and analyzes both streaming and batch data in faster, scalable and more efficient ways. Designed around the Kappa Architecture, Focus™ uses the same code for training, testing and deploying algorithms, and therefore handles streaming and batch data with the same ease and efficiency.

As the term implies, a microservices architecture allows the user to isolate and operate one micro piece of robust software without affecting other pieces. “This is a growing trend in the market : LinkedIn, Google and Amazon are among the companies using this approach, marking a distinct move away from historically monolithic systems,” said Kate Ravanis, Via Science’s Vice President, Client Service and Operations.

Via Science’s industrial clients largely drove the shift to a microservices architecture. Despite similar objectives across clients (e.g., risk assessment, causal inference and anomaly detection), the variation in the types and frequency of data as well as differences in operational deployment required an approach with smaller, more independent modules.

“One microservice of Focus™ is the use of Apache Avro, a data description schema. This allows for easy communication and integration with other data sources regardless of operating systems, which speeds up customer collaboration,” said Jeremy Taylor, Vice President, Lead Scientist of Via Science.

At the core of Focus™ is REFS™, Via Science’s patented machine learning engine that can automatically infer cause and effect relationships in data. Causal inference creates more accurate predictive algorithms and is uniquely capable of addressing diagnostic and simulation problems. The underlying mathematical approach of REFS™ is Bayesian networks, which makes the engine particularly well-suited to handle missing and sparse data: a major problem in real world applications like power grids or in cybersecurity.

Other upgrades in this release include utilizing Docker and EasyBuild to automate Amazon EC2 and on-premise installation, and Spring IO, Hadoop and Flink (to name a few) to streamline data processing and publishing outputs in standardized formats like JSON and REST API for web and mobile access. The final algorithms are lightweight and deployable on a wide range of infrastructures.

Currently, Via Science is using Focus™ to develop a public safety app in collaboration with Rio de Janeiro-based think tank, the Igarapé Institute. The app will generate relative crime rates for neighborhoods in urban areas, starting with Rio. Focus™ analyzed crime and other regional data to generate the ratings, which aim to promote awareness of potential safety concerns for travelers. Via Science and the Igarapé Institute plan to release the beta version of the app during this summer’s Olympic Games in Rio.

Focus Architecture

Focus™ architecture uses a modular approach, allowing us to offer customized solutions.