AI Biweekly: 10 Bits from Mar W 2 – Mar W 3

March 6th – Google Builds AI Drone Image Analysis System for the Pentagon
Google is collaborating with the United States Department of Defense to create an AI system for identifying drone footage. The system will leverage machine learning and computer vision technologies to collect and categorize drone footage more efficiently.

March 7th – Microsoft Announces an AI Platform for Developers in Windows 10
Microsoft is adding a new AI platform to Windows 10 which includes or enables various AI-based applications such as Photos app for intelligent video creation, Hello app for face recognition log in, and voice assistant Cortana for question answering.

March 7th – Microsoft and Intel Enable Deep Learning Inference Vision Processing Units (VPU)
Microsoft announces Intel’s Movidius Myriad X VPU hardware will be included in the Windows ML server. Windows ML can pre-train machine learning data models and process developers’ own deep learning tasks within the OS interface. With the added power of Intel’s VPU, Windows ML will be able to run specific deep neural network AI tasks at higher speeds while using less power.

March 9th – Telenor Enhances AI and IoT Laboratory Research
Telenor has opened a new IoT laboratory research unit in Trondheim, Norway. The company will invest over €5.2 million in the new lab, which will mainly collaborate with its existing AI lab on Narrowband IoT technology research and developer portal enhancement.

March 9th – Waymo and Google Launch an Autonomous Truck Pilot Project in Atlanta
Waymo and Google announce a pilot project for self-driving truck testing in Atlanta, USA. During this project the team will not only test the capabilities of their autonomous truck but also the automated logistics in allocating loads, connecting shippers, factories, and distribution centres.

March 13th – Google Makes Music with Machine Learning
Under its Magenta project, Google introduces its NSynth (Neural Synthesizer), which uses machine learning algorithms and deep neural networks to create new sounds for electronic music. The prototype can create over 100,000 sounds based on only 16 source inputs.

March 14th – Huawei Introduces Blockchain Stress Test Project
Huawei introduces its Caliper Project, a method for testing blockchain stress performance which can help developers and engineers evaluate their blockchain applications in a controlled environment. Performance results may include success rate of the transaction, speed of transaction, and hardware resource performance.

March 14th – Alexa Now Able to Play AI-Generated Songs
Alexa’s latest skill is the ability to play AI-generated songs. The new DeepMusic feature uses a deep recurrent neural network to collect music samples, and leverages an algorithm to create melodies and songs without human input.

March 15th – Microsoft’s AI Translation System Performs at Human Expert Level
Microsoft builds a Chinese to English translation system with natural language processing and machine learning techniques that can translate Mandarin to English at a level matching human translators. Microsoft based its results on blind comparisons by bilingual humans between its system’s outputs and those of human translators.

March 15th – Google Ventures Invests in AI Chip Start-Up Sambanova
Google Ventures’ parent company Alphabet pours US$56 million in Series A funding into startup SambaNova. SambaNova builds computer chipsets and applications for artificial intelligence and data analytics. SambaNova CEO Rodrigo Liang says the company is also interested in public cloud services.