BrainChip Presents at AI Field Day 1

Anil Mankar presented at AI Field Day 1

This Presentation date is November 19, 2020 at 9:00-10:00.

Introducing the BrainChip Akida Event-Domain Neural Processor

Anil Mankar, Co-Founder and Chief Development Officer, introduces the BrainChip Akida ultra low power edge AI solution. He begins with a discussion of the event-based architecture of the Akida technology, and how it addresses the challenges of edge computing. The Neuromorphic design enables distributed computation and event-based processing, communication, and learning. Akida utilizes low-bit precision to reduce memory and bandwidth requirements yet still delivers accuracy in leading models.

Personnel: Anil Mankar

BrainChip Akida Architecture and Evaluation Board Introduction

Watch on Vimeo
Watch on YouTube

Anil Mankar, Co-Founder and Chief Development Officer, provides a tour of the BrainChip Akida NPU architecture and IP solution. He also discusses the Akida software development environment (ADE) and training workflow before turning to a presentation of the first Akida based products. The AKD1000 NSoC is a chip that incorporates an Akida Neuron fabric along with all necessary supporting interfaces and an M-Class CPU. The chip is packaged in an M.2 PCIe form factor evaluation board and enables efficient processing of various sensor modalities, as demonstrated later in this session.

Personnel: Anil Mankar

Neural Networks Optimized to Run on BrainChip Akida Technology

Watch on Vimeo
Watch on YouTube

Anil Mankar, Co-Founder and Chief Development Officer, presents the various neural networks optimized to operate on the BrainChip Akida processor. He begins with auditory classification, discussing keyword spotting in an always on microphone. He then presents visual processing, using person detection as an example. Next he discusses olfactory detection using the Fox 3000 olfactory data set and detecting volatile organic compounds. The next sense is taste classification using an “e-tongue”. Finally he presents somatosensory classification with an example of electric motor ball bearing fault diagnosis.

Personnel: Anil Mankar