7 papers
Large Pre-Training Datasets Don't Always Guarantee Robustness after Fine-Tuning
Jaedong Hwang, Brian Cheung, Zhang-Wei Hong +3
Large-scale pretrained models are widely leveraged as foundations for learning new specialized tasks via fine-tuning, with the goal of maintaining the general performance of the mo…
Breaking Neural Network Scaling Laws with Modularity
Akhilan Boopathy, Sunshine Jiang, William Yue +3
Modular neural networks outperform nonmodular neural networks on tasks ranging from visual question answering to robotics. These performance improvements are thought to be due to m…
Permutation Invariant Learning with High-Dimensional Particle Filters
Akhilan Boopathy, Aneesh Muppidi, Peggy Yang +3
Sequential learning in deep models often suffers from challenges such as catastrophic forgetting and loss of plasticity, largely due to the permutation dependence of gradient-based…
Unified Neural Network Scaling Laws and Scale-time Equivalence
Akhilan Boopathy, Ila Fiete
As neural networks continue to grow in size but datasets might not, it is vital to understand how much performance improvement can be expected: is it more important to scale networ…
Grid Cell-Inspired Fragmentation and Recall for Efficient Map Building
Jaedong Hwang, Zhang-Wei Hong, Eric Chen +3
Animals and robots navigate through environments by building and refining maps of space. These maps enable functions including navigation back to home, planning, search and foragin…
Towards Exact Computation of Inductive Bias
Akhilan Boopathy, William Yue, Jaedong Hwang +2
Much research in machine learning involves finding appropriate inductive biases (e.g. convolutional neural networks, momentum-based optimizers, transformers) to promote generalizat…