| Resumo: | <p>The Bodleian Libraries have recently launched the ORA (Oxford Research Archive) collection on Artificial Intelligence and Machine Learning. This portal collects the AI-related research output from Oxford authors across all disciplines in ORA.</p>
<p>As part of the launch of this new resource, an event with knowledgeable researchers in the field of AI and Machine Learning at Oxford was held, consisting of a series of short talks, poster display, Q&A, and networking.</p>
<br>
<p>A breakdown of the event as follows:</p>
<br>
<p><b>Introduction:</b></p>
<p>Alessandra Vetrugno, Lead Librarian for Physical and Applied Sciences, Bodleian Libraries;</p>
<p>Jason Partridge, Open Access Service Manager, Bodleian Libraries;</p>
<p>Rachel Scanlon, Subject Librarian for Physical and Applied Sciences, Bodleian Libraries</p>
<br>
<p><b>Lightning talks:</b></p>
<p><b>Transfer learning for heterocycle synthesis prediction</b></p>
<p>Ewa Wieczorek, DPhil student in the Department of Chemistry</p>
<p>While heterocycles play a central role in medicinal chemistry, predicting synthetic pathways towards them remains challenging for computational tools. This talk shows how the use of various transfer learning strategies with a transformer model can lead to increased accuracy of ring-formation prediction.</p>
<br>
<p><b>Machine learning-based potentials for modelling chemical reactions in the gas phase and solution</b></p>
<p>Veronika Juraskova, Postdoctoral researcher in the Department of Chemistry</p>
<p>This talk will demonstrate how machine learning-based potentials accelerate the accurate modelling of chemical reactions.</p>
<br>
<p><b>Learning to Adapt: Rational Personalization of Cancer Treatment Schedules using Deep Reinforcement Learning</b></p>
<p>Kit Gallagher, DPhil student in the Mathematical Institute</p>
<p>Proposing the application of deep reinforcement learning to guide adaptive drug scheduling in oncology, and demonstrating that this framework can generate rational treatment schedules that double the time to relapse attained by current adaptive protocols.</p>
<br>
<p><b>Simulation and analysis of highly polygenic traits</b></p>
<p>Daiki Tagami, Dphil student in the Department of Statistics</p>
<p>Using machine learning to analyze highly polygenic traits from the human whole-genome sequencing data.</p>
<br>
<p><b>AI & Elections in Africa</b></p>
<p>Michael Collyer, DPhil student at the Oxford Internet Institute</p>
<p>This talk will highlight some of the opportunities and challenges regarding the use of AI in elections with a focus on Africa.</p>
<br>
<p><b>International governance of civilian AI via jurisdictional certification</b></p>
<p>Benjamin Harack, DPhil student in the Department of Politics and International Relations and DPhil Affiliate at the Oxford Martin AI Governance Initiative</p>
<p>AI governance must include standardized international regulation, and one way to accomplish that goal is through a certification regime similar to what already exists for civil aviation, shipping, and banking.</p>
<br>
<p><b>Posters:</b></p>
<p><b>Using AI-generated speech to inform historical sound change (with specific reference to Old English vowel breaking)</b></p>
<p>Jonathan Wei, DPhil student in the Department of Linguistics, Philology and Phonetics</p>
<br>
<p><b>A Reinforcement Learning approach to Hamiltonian Eigenvalue Solving</b></p>
<p>Oliver Chapman, DPhil student in the Department of Chemistry</p>
<br>
<p><b>Trustworthiness Auditing for Artificial Intelligence Systems</b></p>
<p>Kaivalya Rawal, Postdoctoral researcher at the Oxford Internet Institute</p>
|
|---|