The following schedule is a work in progress. Once we have the full seminar details including title, abstract and speaker bio, we will publish these talks under upcoming seminars.
| Date | Speaker | Title | Abstract |
|---|---|---|---|
| 2026-03-12 11:00 | Dr Nhan Ly-Trong | MLFormer: A Transformer-based approach for phylogenetic tree search | Phylogenomics plays a vital role in elucidating evolutionary histories of species on Earth, ranging from early life forms billions of years ago to kangaroos, gum trees, and the recent emergence of the SARS-CoV-2 virus causing the COVID-19 pandemic. Widely used phylogenetic inference tools, such as IQ-TREE, RAxML, and PHYML, employ heuristic tree search algorithms to construct phylogenetic trees that maximize the likelihood of observed genetic sequence alignments. However, these tree search methods are computationally intensive and often prone to local optima. To address these challenges, we present MLFormer, a Transformer-based approach to propose tree rearrangements for tree search. Experimental results show that MLFormer helps tree search escape local optima and obtains reasonable accuracy while being significantly faster than IQ-TREE. To reconstruct 1,000 phylogenetic trees from 20-taxon alignments, IQ-TREE required a total of 21 hours while MLFormer took only 3 minutes, a speedup of over 400 times. |
| 2026-03-19 11:00 | Gepeng Ji | ||
| 2026-03-20 11:00 | Yonatan Gideoni | What matters for code evolution? | Recent works like AlphaEvolve demonstrated LLM-based search pipelines solving various problems by finding computer programs for them. Most code evolution pipelines consist of many design choices which are not thoroughly ablated, making it difficult to understand their importance. Testing two simple baselines, we find that much simpler methods can match or exceed purpose-built code evolution pipelines across three domains, each having different constraints. In this talk I will discuss why simple baselines do so well, shortcomings in how code evolution is evaluated and applied, and broader implications for the field. |
| 2026-03-26 11:00 | Zora Zhuang | Machine learning for flow cytometry: the present, the opportunities, the challenges | Flow cytometry characterises cells through noisy, partial observations of the cell surface, generating medium-resolution cellular features in an affordable manner. More recently, deep learning methods have been deployed to link cellular patterns in flow cytometry to complex patient conditions, to various degrees of success. In this talk, we'll discuss both the potentials and the limitations of flow cytometry data, the challenges they present to ML based methods, and a new approach that could utilise flow data more effectively. |
| 2026-04-02 11:00 | |||
| 2026-04-09 11:00 | |||
| 2026-04-16 11:00 | |||
| 2026-04-23 11:00 | |||
| 2026-04-30 11:00 | |||
| 2026-05-07 11:00 | |||
| 2026-05-14 11:00 | |||
| 2026-05-21 11:00 | |||
| 2026-05-28 11:00 | |||
| 2026-06-04 11:00 | |||
| 2026-06-11 11:00 | |||
| 2026-06-18 11:00 | |||
| 2026-06-25 11:00 | |||
| 2026-07-02 11:00 | |||
| 2026-07-09 11:00 | |||
| 2026-07-16 11:00 | |||
| 2026-07-23 11:00 | |||
| 2026-07-30 11:00 | |||
| 2026-08-06 11:00 | |||
| 2026-08-13 11:00 | |||
| 2026-08-20 11:00 | |||
| 2026-08-27 11:00 | |||
| 2026-09-03 11:00 | |||
| 2026-09-10 11:00 | |||
| 2026-09-17 11:00 | |||
| 2026-09-24 11:00 | |||
| 2026-10-01 11:00 | |||
| 2026-10-08 11:00 | |||
| 2026-10-15 11:00 | |||
| 2026-10-22 11:00 | |||
| 2026-10-29 11:00 | |||
| 2026-11-05 11:00 | |||
| 2026-11-12 11:00 | |||
| 2026-11-19 11:00 | |||
| 2026-11-26 11:00 | |||
| 2026-12-03 11:00 |