Jeff Brower

Background

About Jeff Brower

Technical and business expert in deep learning, telecom, HPC, supercomputing, signal and image processing and algorithms, and embedded systems. Responsible for deliveries and deployments in a wide range of areas, including avionics, telecom, speech processing, lawful intercept, robotics, and deep learning. For demo Rar packages and Docker containers see https://github.com/signalogic/SigSRF_SDK and https://hub.docker.com/repositories/signalogic Currently working on real-time ASR (automatic speech recognition) in small form-factor applications. To handle multiple concurrent RTP packet streams, unify stream conversations, and recognize/transcribe conversational speech in pico ITX form-factors is extremely challenging. My research includes alternative training methods based on CAM,"memory topography", sparse vectors, and other AI language model methods more likely used in the human brain than back propagation, gradient descent, and other complex methods that demand exact, error-free computation, requiring Watts of energy consumption far exceeding the 40 to 60 Watts consumed by the brain. Such inefficient, energy sucking methods have not evolved in nature, nor are likely to.