6 citations · 12 across the 18 of their papers we have counts for
6 papers · 1 filter
Represented Is Not Computed: A Causal Test of Candidate Algorithmic Intermediates in a Transformer
Ishita Darade, Sushrut Thorat
Structured prompts require integrating components according to task-relevant relations. How a network implements this integration is often hard to judge in language or vision, wher…
Path Integration and Object-Location Binding Emerge in an Action-Conditioned Predictive Sequence Network
Linda Ariel Ventura, Victoria Bosch, Tim C Kietzmann +1
Adaptive cognition requires structured internal models of objects and their relations. Predictive neural networks are often proposed to learn such world models, but how these are i…
Brain-language fusion enables interactive neural readout and in-silico experimentation
Victoria Bosch, Daniel Anthes, Adrien Doerig +3
Large language models (LLMs) have revolutionized human-machine interaction, and have been extended by embedding diverse modalities such as images into a shared language space. Yet,…
Adopting a human developmental visual diet yields robust, shape-based AI vision
Zejin Lu, Sushrut Thorat, Radoslaw M Cichy +1
Despite years of research and the dramatic scaling of artificial intelligence (AI) systems, a striking misalignment between artificial and human vision persists. Contrary to humans…
Diagnosing Catastrophe: Large parts of accuracy loss in continual learning can be accounted for by readout misalignment
Daniel Anthes, Sushrut Thorat, Peter König +1
Unlike primates, training artificial neural networks on changing data distributions leads to a rapid decrease in performance on old tasks. This phenomenon is commonly referred to a…
Keep Moving: identifying task-relevant subspaces to maximise plasticity for newly learned tasks
Daniel Anthes, Sushrut Thorat, Peter König +1
Continual learning algorithms strive to acquire new knowledge while preserving prior information. Often, these algorithms emphasise stability and restrict network updates upon lear…