Mechanistic Mode Connectivity

11/15/2022
by   Ekdeep Singh Lubana, et al.
0

Neural networks are known to be biased towards learning mechanisms that help identify spurious attributes, yielding features that do not generalize well under distribution shifts. To understand and address this limitation, we study the geometry of neural network loss landscapes through the lens of mode connectivity, the observation that minimizers of neural networks are connected via simple paths of low loss. Our work addresses two questions: (i) do minimizers that encode dissimilar mechanisms connect via simple paths of low loss? (ii) can fine-tuning a pretrained model help switch between such minimizers? We define a notion of mechanistic similarity and demonstrate that lack of linear connectivity between two minimizers implies the corresponding models use dissimilar mechanisms for making their predictions. This property helps us demonstrate that naïve fine-tuning can fail to eliminate a model's reliance on spurious attributes. We thus propose a method for altering a model's mechanisms, named connectivity-based fine-tuning, and validate its usefulness by inducing models invariant to spurious attributes.

READ FULL TEXT

page 17

page 18

research
08/24/2023

Geodesic Mode Connectivity

Mode connectivity is a phenomenon where trained models are connected by ...
research
11/24/2022

Prototypical Fine-tuning: Towards Robust Performance Under Varying Data Sizes

In this paper, we move towards combining large parametric models with no...
research
09/21/2021

Stepmothers are mean and academics are pretentious: What do pretrained language models learn about you?

In this paper, we investigate what types of stereotypical information ar...
research
05/24/2022

Linear Connectivity Reveals Generalization Strategies

It is widely accepted in the mode connectivity literature that when two ...
research
08/22/2023

Mode Combinability: Exploring Convex Combinations of Permutation Aligned Models

We explore element-wise convex combinations of two permutation-aligned n...
research
05/18/2023

Mode Connectivity in Auction Design

Optimal auction design is a fundamental problem in algorithmic game theo...
research
09/28/2022

Topological descriptors of the parameter region of multistationarity: deciding upon connectivity

Switch-like responses arising from bistability have been linked to cell ...

Please sign up or login with your details

Forgot password? Click here to reset