AIConfPaper
← Search

Hamza Merzic

5 accepted papers

2024

Bad Students Make Great Teachers: Active Learning Accelerates Large-Scale Visual Understanding

Talfan Evans*, Shreya Pathak, Hamza Merzic, Jonathan Richard Schwarz, Ryutaro Tanno, Olivier Henaff*
ECCV 2024poster

"Power-law scaling indicates that large-scale training with uniform sampling is prohibitively slow. Active learning methods aim to increase data efficiency by prioritizing learning on the most relevant examples. Despite their appeal, these methods have yet to be widely adopted since no one algorithm…

Cited by 13Source ↗PDF ↗Scholar ↗
2024

Data curation via joint example selection further accelerates multimodal learning

Talfan Evans, Nikhil Parthasarathy, Hamza Merzic, Olivier J Henaff
NeurIPS 2024spotlight

Data curation is an essential component of large-scale pretraining. In this work, we demonstrate that jointly prioritizing batches of data is more effective for learning than selecting examples independently. Multimodal contrastive objectives expose the dependencies between data and thus naturally y…

Cited by 18Source ↗PDF ↗Scholar ↗
2021

Grounded Language Learning Fast and Slow

Felix Hill, Olivier Tieleman, Tamara von Glehn, Nathaniel Wong, Hamza Merzic, Stephen Clark
ICLR 2021spotlight

Recent work has shown that large text-based neural language models acquire a surprising propensity for one-shot learning. Here, we show that an agent situated in a simulated 3D world, and endowed with a novel dual-coding external memory, can exhibit similar one-shot word learning when trained with c…

Cited by 85Source ↗PDF ↗code ↗Scholar ↗
2020

Probing Emergent Semantics in Predictive Agents via Question Answering

Abhishek Das, Federico Carnevale, Hamza Merzic, Laura Rimell, Rosalia Schneider, Josh Abramson +5
ICML 2020poster

Recent work has shown how predictive modeling can endow agents with rich knowledge of their surroundings, improving their ability to act in complex environments. We propose question-answering as a general paradigm to decode and understand the representations that such agents develop, applying our me…

Cited by 22Source ↗PDF ↗Scholar ↗
2019

Shaping Belief States with Generative Environment Models for RL

Karol Gregor, Danilo Jimenez Rezende, Frederic Besse, Yan Wu, Hamza Merzic, Aaron van den Oord
NeurIPS 2019poster

When agents interact with a complex environment, they must form and maintain beliefs about the relevant aspects of that environment. We propose a way to efficiently train expressive generative models in complex environments. We show that a predictive algorithm with an expressive generative model can…

Cited by 127Source ↗PDF ↗Scholar ↗