IFML Seminar
IFML Seminar: 11/07/25 - Model Self-improvement via Optimal Retraining
Abstract: Retraining a model using its own predictions together with the original, potentially noisy labels is a well-known strategy for improving the model’s performance. While prior works have demonstrated the benefits of specific heuristic...
Upcoming Events
- November712:15 - 1:15pm
IFML Seminar
Event DetailsAbstract: Retraining a model using its own predictions together with the original, potentially noisy labels is a well-known strategy for...
- November1412:15 - 1:15pm
IFML Seminar
Event DetailsAbstract: Diffusion language models (DLMs) represent a nascent but promising alternative to GPT-style autoregressive (AR) language models: as opposed to...
- November2112:15 - 1:15pm
IFML Seminar
Event DetailsAbstract: Intelligence often emerges through interaction and competition. Likewise, advanced AI algorithms often rely on competing learning objectives.…
Past Events
- November412:15 - 1 pm
IFML Seminar
Event DetailsA central problem in machine learning is as follows: How should we train models using data generated from a collection...
- October2812:15 - 1 pm
IFML Seminar
Event DetailsAbstract: Although Machine learning (ML) algorithms have recently made a huge impact on medical imaging, their development and deployment for...
- October2112:15 - 1 pm
IFML Seminar
Event DetailsRepresentation learning has been widely used in many applications. In this talk, I will present our work which uncovers when...
- October1412:15 - 1 pm
IFML Seminar
Event DetailsWhen Is Partially Observable Reinforcement Learning Not Scary? Abstract: Partially observability is ubiquitous in applications of Reinforcement Learning (RL…
- September2312:15 - 1 pm
ML+ X Seminar
Event DetailsAbstract: In this talk I will present applications and new methods for decomposing higher-order moment tensors into appropriate low-rank representations...
- September1612 - 1 pm
IFML Seminar
Event DetailsTitle: A Data-Centric View on Reliable Generalization Abstract: Researchers have proposed many methods to make neural networks more reliable under...
