Biological Computation in Machine Learning
William Poole is a PhD. Candidate at Caltech working on Bridging Computation with Synthetic Biology. William’s research connects Machine Learning / AI theory to our understanding of real and synthetic biochemical networks. In the longer term, William would like to use his theoretical and computational skills to further the burgeoning field of synthetic biology and develop technologies around sustainability and climate change.
From identifying images to playing video games to understanding genomics, artificial intelligence (AI) and in particular machine learning (ML) have allowed computers to perform tasks previously thought to be distinctly human. Moreover, at the cutting edge of scientific research into systems neuroscience and systems biology, ML has become an invaluable tool for extending human intelligence to make sense of immensely complicated systems and data. However, there is another side to this story. Many of the most successful ideas in ML – including deep learning, convolutional networks, and reinforcement learning – were inspired by how brains actually seem to work. More recently, connections have been made between ML and other scales of biological computation, such as the biochemical networks in single cells. This talk will anecdotally introduce the cross-pollination between neuroscience, biology, and AI. From this historical perspective, I will suggest that ML’s general-purpose usefulness is no surprise and that biology still has the potential to provide inspiration for the next generation of AI algorithms.