The Research and Applied AI Summit (RAAIS) is a community for entrepreneurs and researchers who accelerate the science and applications of AI technology. In the lead up to our 5th annual event on June 28th 2019 in London, we’re running a series of speaker profiles to shed more light on what you can expect to learn on the day!

andrew trask

One of the most pertinent topics in AI today is that of data privacy and model ownership. Significant value has accrued to large organisations as they understandably act as centers of gravity for talent and capital. However, for the long-term benefits of AI to be enjoyed by everyone, its important to work on distributing tools, data and ownership broadly. To explore this critical thread of work, we’re excited to welcome Andrew Trask to RAAIS 2019! Andrew is a PhD Student at the University of Oxford, a Research Scientist at DeepMind in London and is the lead of OpenMined. His work focuses on Privacy and Deep Learning. In addition to his research, Andrew is the author of Grokking Deep Learning, a Manning Publications introductory book which has sold over 10,000 copies (it’s not even a year old!). Andrew is also an instructor in Udacity’s Deep Learning Nanodegree, and the author of a popular machine learning blog (where I first connected with him in early 2017).

For those of you who are actively involved in the open source machine learning community, you’ll probably know Andrew from the OpenMined project that he kickstarted in 2018. OpenMined is now a community of 3,500 machine learning researchers, practitioners, and enthusiasts building open source tools for safe, privacy-preserving Deep Learning.

In his session at RAAIS 2019, Andrew will motivate the need for research and real-world applications of privacy-preserving AI systems. He will unpack the key components to the OpenMined system, which includes multiparty computation, differential privacy, federated learning, homomorphic encryption and novel governance mechanisms. Taken together, the OpenMined system allows an AI model to be governed by multiple owners and trained securely on an unseen, distributed dataset. This contrasts with industry standard tools for AI that have been designed under the assumption that data is centralized into a single compute cluster, the cluster exists in a secure cloud, and the resulting models will be owned by a central authority.

Prior to starting his PhD at Oxford, Andrew lead product analytics at Digital Reasoning, which delivers enterprise A.I. solutions to Hedge Funds, Investment Banks, Healthcare Networks, and Government Intelligence clients. While at Digital Reasoning, he also trained the world's largest neural network, a record published as a part of his work on neural word embeddings at the International Conference on Machine Learning in 2015.

You can find Andrew on GitHub here, on his blog here, and you can read up on his research here.