According to Analytics Insight, Berkeley Artificial Intelligence Research(BAIR) Lab is one of the leading research labs in the world in the field of AI. BAIR works with the University of California, Berkeley to research in the field of Computer vision, machine learning, Natural Language Processing(NLP), and Robotics. BAIR has a team of 50 faculty and over 300 students researching about latest innovations and breakthroughs in artificial intelligence and related domain.
BAIR was established in 2017 with the purpose of creating a dedicated facility under University of California, to accelerate research in AI and robotics. With advancement , it has now moved from research to production. Microsoft Azure and Amazon Web Services have partnered with BAIR to provide computational resources and solve real world business problems with the help of AI. NVIDIA has provided BAIR with latest high supercomputer that was used for reinforcement learning research on robotic arm.
BAIR has its own blog site. It is one of the best websites to follow if you want to stay up-to-date with latest researches in AI. BAIR blogs are detailed information about a research done in their lab and their results.
Convolutional Architecture for Fast Feature Embedding is a deep learning framework that BAIR labs developed. It was designed to enhance the speed and modularity of the framework. It is open-source and was written in C++, which has a python interface. CAFFE features are:
- Optimal CPU and GPU usage.
- High-speed image processing.
- Code backed by a large community of contributors.
Facebook has made more progress and released CAFFE 2.0, which was then merged into PyTorch(an open-source machine learning library)
BAIR started in 2017 and the initial project they did was to teach a robotic arm to learn to pick up different objects with the help of reinforcement learnining and no pretrained data. Nvidia supported this project by providing high end GPU for processing visuals and converting them as input to AI.
BAIR has open sourced their latest driving dataset which contains 100K videos of 40 seconds which is used by the AI community to train self driving car algorithm. It is state-of-the-art dataset with videos having real life driving conditions including rain, fog, night situations.
Berkeley’s initiative in focus towards AI will definitely help in getting a breakthrough for human technology. For staying upto date with cutting edge research following BAIR labs research work will prove to be very beneficial.