The company develops artificial intelligence (AI) software purpose-built for engineers, scientists, and researchers that is designed to enable them to “innovate and make discoveries faster.”
“Part of Noble’s mission is building AI that’s accessible to engineers, scientists, and researchers, anytime and anywhere, without needing to learn or re-skill into computer science or AI theory,” says Dr. Matthew C. Levy, Founder and CEO of Noble.AI. “The reason why we’re making this symbolic contribution open source is so people have greater access to tools amenable to R&D problems.”
The company’s specific contribution helps to augment the “sparse matrix” capabilities of TensorFlow. Often, says the company, matrices represent mathematical operations that need to be performed on input data, such as in calculating the temporal derivative of time-series data.
In many common physics and R&D scenarios these matrices can be sparsely populated such that a tiny fraction – often less than one percent – of all elements in the matrix are non-zero. In this setting, storing the entire matrix in a computer’s memory is cumbersome and often impossible all together at R&D industrial scale. In these cases, it often becomes advantageous to use sparse matrix operations.
For more information and usage particulars, visit the Noble.AI open-source GitHub.
Open-source library for rapid prototyping of quantum ML models
Google neural network library ‘opens the black box’ of deep learning
Wave Computing to donate training scheme to TensorFlow
Amazon open sources AI project to accelerate ML on edge devices
CEVA supports TensorFlow Lite for Microcontrollers