December 11th – MIT Scientists Teach AI to Have Emotions
Researchers at the MIT Media Lab partner with McKinsey to create a machine learning program, which can recognize and predict emotional responses from movies. The program studies facial expressions and other details to understand the level of emotional response a scene elicits from viewers. It successfully learns to recognize the emotional content of videos and separates the positive from negative.
December 11th – AI Company Stem Works with Mitsui to Build the First Energy Storage in Japan
After the Fukushima Daiichi nuclear accident in 2011, the Japanese government decides to redesign the country’s 300-GW of electricity grids. This huge workload provides an opportunity for AI companies. Stem Inc, an international provider of AI-driven energy storage services is working with Mitsui & Co., Ltd. to build the first energy storage unit in Japan. Stem’s AI will help reduce energy costs by providing better control and efficiency.
December 12th – New York City Implements Expert Team to Correct Algorithm Biases
New York City plans to implement a dedicated team of experts that will monitor the fair use of algorithms in municipal agencies. The team is responsible for issuing a report that describes the procedure made by AI decision systems. The act primarily targets biases inherent in training data. The whole process is designed to ensure “fairness, accountability, and transparency”. According to news sources, the report will be made available to the public in 18 months.
December 13th – eBay Acquires E-Commerce Analytics Company Terapeak
EBay acquires Toronto-based e-commerce analytics startup Terapeak and aims to integrate it into its Seller Hub. Based on transaction data collected, Terapeak provides sellers with suggestions of product listing, inventory management, and pricing. With insights on the market, Terapeak can help mitigate risk and aid seller decision-making process. As the result of this acquisition, Terapeak with work with eBay exclusively with deal closing before the end of 2017.
December 15th – Microsoft is Working on AI & IoT Infrastructural Projects in India
Microsoft is implementing AI & IoT projects in different part of India. Notable examples include *Microsoft FarmBeats *focusing on data collection for agriculture, an AI-based sowing APP for farmers, smart street lighting in the city of Jaipur, and AI-powered interactive cane to help the visually-impaired. These projects demonstrate the use of AI in Asia’s most populous country.
December 15th – NASA Uses Google Neural Network to Detect Exoplanet
NASA has detected a star like our sun using a Google AI trained by data from the Kepler Space Telescope. The network was trained with 15,000 labeled data from the Kepler exoplanet catalog and was able to learn to detect weaker signals. NASA plans to use the neural network to examine more incoming data from the Kepler telescope which consists of 150,000 more stars.
December 15th – Amazon Establishes the ML Solutions Lab to Guide AWS Clients
In order to remain competitive with Microsoft and Google in the cloud computing business, Amazon establishes the ML Solutions Lab, which allows company experts to guide clients in developing products through AWS. Based on numbers from the latest quarter financial report, AWS generated US $4.6 billion in revenue, demonstrating that AWS has become one of Amazon’s most profitable business unit.
December 17th – Microsoft Invests $50 million in AI for Earth Project
Microsoft will invest US $50 million commitment in the AI for Earth program over the next five years. The company will grant universities, companies, and other organizations free use of Microsoft’s AI technology including Azure cloud and mapping tools. Companies working on climate change, water, agriculture, and biodiversity are eligible to apply. One example is Project Premonition which focuses on detecting early pathogens that can cause diseases.
December 19th – MIT and Michigan State University Introduce Auto-Tuned Models (ATM)
MIT’s Laboratory for Information and Decision Systems (LIDS) and Michigan State University introduce a new system that automates machine learning model selection. The new system is called ATM (Auto-Tuned Models) and runs in the cloud. It basically searches over a large number of modeling options with high-throughput and finds the best modeling technique for a given problem. It also tunes hyperparameters. It either outperforms human data scientists or comes close in terms of choosing the right solution, but it is much faster than a human.
December 20th – India’s NASSCOM and China’s Dalian Government Jointly Launch an IoT and AI Platform
India’s National Association of Software and Services Companies (NASSCOM) and China’s Dalian Municipal People’s Government announce the joint investment for an AI and IoT collaborative platform. The partnership will facilitate technical exchanges between Indian and Chinese companies. The Sino-Indian Digital Collaboration Plaza (SIDCOP) platform will be able to operate both online and offline environment.
Industry Analyst: Paul Fan | Editor: Meghan Han