Abstract:

Electrocardiogram (ECG) is widely considered the primary test for evaluating cardiovascular diseases. However, the use of AI models to advance these medical practices and learn new clinical insights from ECGs remains largely unexplored. Utilising a data set of 2.3 million ECGs collected from patients with 7 years follow-up, we developed a DNN model with state-of-the-art granularity for the interpretable diagnosis of cardiac abnormalities, gender identification, and hypertension screening solely from ECGs, which are then used to stratify the risk of mortality. Our model demonstrated cardiologist-level accuracy in interpretable cardiac diagnosis, and the potential to facilitate clinical knowledge discovery for gender and hypertension detection which are not readily available. In addition, we explored the design of optimal DNN models through of a novel Neural Architecture Search (NAS) approach, which was able to find networks outperformed the state-of-the-art models with fewer than 5% parameters.



Bio:

Dr. Lei Lu is a Lecturer in Health Data Science and AI at King’s College London, and a Visiting Research Fellow at the Institute of Biomedical Engineering, University of Oxford. Lei obtained his PhD from the Harbin Institute of Technology in China, complemented by two-year visiting research at the University of British Columbia in Canada. Upon completing his PhD study, Lei had his postdoctoral research at the University of Melbourne in Australia. Subsequently, he joined the Institute of Biomedical Engineering at University of Oxford as a Senior Research Associate. Lei’s work focuses on clinical machine learning to advance healthcare outcomes. Lei is actively engaged in a range of academic roles, including invited speaker at the IET Annual Healthcare Lecture and the IEEE-EMBS Symposium on MDBS. He also served as conference session chair, workshop committee, and guest editor for IJCAI, CIKM, ICRA, and IEEE JBHI. He received the IET J.A. Lodge Award in 2021, which is presented annually to one early-career researcher with distinction.

This talk will examine the emergence of two pre-Islamic Arabian theonyms, Allāh and Allāt, tracking their development across the epigraphic record spanning more than a millennium before the rise of Islam and across the Peninsula.

As a Lunch Table event, members of the AWRC are invited to join Ahmad for lunch in Hall at 12.30. The talk, beginning at 1.15 in the Levett Room, is catered with tea/coffee and cakes (all welcome).

The AWRC Maritime Day will include two 20-minute presentations: one by Linda Hulin, who will speak on 'The Practical Mariner Project', which uses landscape archaeology methods to explore the Bronze Age Mediterranean (ca. 1750-1500 BCE) from the point of view of the sailors, and one by Damian Robinson, who will speak on current underwater archaeology research in 'The Central Harbour of Thonis-Heracleion' project. Presentations in the Levett Room (catered with nibbles and soft drinks) will be followed by a punting expedition up the River Cherwell to the Victoria Arms. RSVP Christoph for punt reservations by 20th May ancient.world@wolfson.ox.ac.

Classical literature tells us the story of soldiers coming back from Alexander’s conquest of the Achaemenid Empire carrying a fortune in gold and gems. Examination of coin production and analyses of its metal can teach us more about where the gold of Alexander came from and where it travelled after the conquest.

As a Lunch Table event, members of the Cluster are invited to join Frédérique for lunch in Hall at 12.30. The talk, beginning at 1.15 in the Levett Room, is catered with tea/coffee and cakes (all welcome).

The four speakers in this event each have 5 slides and 5 minutes to talk about aspects of their research, with each talk followed by 15 minutes of conversation. The speakers are: Prerita Govil (Classical Indian Religion/Comparative Philosophy), Joel Bellviure Perez (Classical Archaeology), Ellen Sharman (History/Early Modern Reception), Alessia Zubani (History of Science/Late Antiquity).



The event is catered with wine, soft drinks and nibbles.

The Indo-European language family comprises ten primary branches (Anatolian, Tocharian, Italic, Celtic, Germanic, Balto-Slavic, Albanian, Greek, Armenian, and Indo-Iranian). The academic study of these languages together as a group may be said to begin in 1816, and the method of linguistic reconstruction still in use crystallised in the second half of the 19th century. The 20th and early 21st centuries saw the discovery of the Tocharian and Anatolian branches, as well as a number of changes and refinements in theory. What is the situation of these studies today, and where are they (~should they be) going?



As a Lunch Table event, members of the Cluster are invited to join Tim for lunch in Hall at 12.30. The talk, beginning at 1.15 in the Levett Room, is catered with tea/coffee and cakes (all welcome).

The spectacular ‘Main Deposit’ excavated by James Quibell and Frederick Green at Hierakonpolis in 1897–98 included hundreds of fragments of human statuettes carved from hippopotamus and elephant ivory. The cache represents men, women and children in a variety of poses and costumes, ranging from complete examples to the detached heads, arms, legs, feet and bases of many others. Debate continues concerning the date of their manufacture and the reason for their deposition. I will present a new study of the corpus, relating the Hierakonpolis pieces to comparative material from deposits found at other sites across Egypt. I also challenge the standard interpretation of such deposits as discarded temple offerings and propose an alternative explanation of the contexts in which they should be understood.



As a Lunch Table event, members of the Cluster are invited to join Liam for lunch in Hall at 12.30. The talk, beginning at 1.15 in the Levett Room, is catered with tea/coffee and cakes (all welcome).

Bio:

Jakob Foerster started as an Associate Professor at the department of engineering science at the University of Oxford in the fall of 2021. During his PhD at Oxford he helped bring deep multi-agent reinforcement learning to the forefront of AI research and interned at Google Brain, OpenAI, and DeepMind. After his PhD he worked as a research scientist at Facebook AI Research in California, where he continued doing foundational work. He was the lead organizer of the first Emergent Communication workshop at NeurIPS in 2017, which he has helped organize ever since and was awarded a prestigious CIFAR AI chair in 2019.





Abstract:

In general-sum games, the interaction of self-interested learning agents commonly leads to collectively worst-case outcomes, such as defect-defect in the iterated prisoner's dilemma (IPD). To overcome this, some methods, such as Learning with Opponent-Learning Awareness (LOLA), shape their opponents' learning process. However, these methods are myopic since only a small number of steps can be anticipated, are asymmetric since they treat other agents as naive learners, and require the use of higher-order derivatives, which are calculated through white-box access to an opponent's differentiable learning algorithm. In this talk I will first introduce Model-Free Opponent Shaping (M-FOS), which overcomes all of these limitations. M-FOS learns in a meta-game in which each meta-step is an episode of the underlying (``inner'') game. The meta-state consists of the inner policies, and the meta-policy produces a new inner policy to be used in the next episode. M-FOS then uses generic model-free optimisation methods to learn meta-policies that accomplish long-horizon opponent shaping. I will finish off the talk with our recent results for adversarial (or cooperative) cheap-talk: How can agents interfere with (or support) the learning process of other agents without being able to act in the environment?

Learn about the future of biomedical imaging with Professor Konstantinos Kamnitsas and the Wolfson Engineering Society!