Tech giants including Google and Intel turn to Cambridge deep learning ace
Global technology giants have turned to a Cambridge UK deep learning pioneer to help steer a world first AI and ML collaboration.
myrtle.ai has assembled a globally renowned team of experts with expertise in producing low power inference circuits and already works with quoted businesses on both sides of the Atlantic.
Now it has been chosen to develop a Speech Recognition benchmark for MLPerf – a new Machine Learning (ML) benchmarking competition backed by Google, Baidu, Intel and AMD.
MLPerf, a collaboration of tech giants and researchers from numerous universities including Harvard, Stanford and the University of California Berkeley, is aspiring to drive progress in ML by developing a suite of fair and reliable benchmarks for emerging artificial intelligence hardware and software platforms.
myrtle.ai has been selected to provide the computer code that will be the benchmark standard for the Speech Recognition division. The code is a new implementation of two AI models known as DeepSpeech 1 and DeepSpeech 2, building on models originally developed by Baidu.
CEO Peter Baldwin (pictured) said: “We are honoured to be providing the reference implementations for the Speech to Text category of MLPerf. Myrtle has a world-class machine learning group and we are pleased to be able to provide the code as open source so that everyone can benefit from it.”
Baldwin says this is the first time the AI community has come together to try to develop a series of reliable, transparent and vendor-neutral ML benchmarks to highlight performance differences between different ML algorithms and cloud configurations.
The new benchmarking suite will be used to test and measure training speeds and inference times for a range of ML tasks.
myrtle ai’s Speech Recognition benchmark is based on proven experience in this field. Its core R & D team has speeded up Mozilla’s DeepSpeech implementations 100-fold when training on Librispeech, demonstrating their practical experience of training and deploying AI and ML algorithms.
Myrtle was founded to develop software and services for public and private data centres. Originally specialising in image processing and large scale simulation Myrtle helped produce computer generated content for over 20 major Hollywood blockbusters.
Clients have included NYSE and NASDAQ listed companies in LA, Vancouver and London as well as a major automotive OEM and a government department.
The company is currently targeting its technology at inference workloads in data centres and is involved in a major collaboration to address the safety and verification challenges that currently preventing sophisticated deep learning networks being used in road vehicles.
Peter Baldwin has run Myrtle since founding it. He has a pure mathematics PhD from Cambridge University and a special interest in the mathematical foundations of deep learning.
He wrote and licensed Myrtle’s first commercially available software: a suite of simulation tools that ran at huge scale in private data centres. The first commercial software he ever wrote was used to help produce the chocolate river in Tim Burton’s film ‘Charlie and the Chocolate Factory’.
The celebrated founder of Cambridge Angels, Robert Sansom, is a director while another angel and entrepreneur Robert Swann, who was a first mover in the enterprise, is also on the board; two other companies that benefit from his advice are Audio Analytic and Undo Software – both global leaders in their fields of tech specialism.
From world-changing self-driving car projects to reducing the power consumption of global data centres, myrtle.ai strives to use its engineering expertise to transform the world today by making tomorrow’s AI run with unsurpassed efficiency on low power hardware.