Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Madan, Spandan, Xiao, Will, Cao, Mingran, Pfister, Hanspeter, Livingstone, Margaret, Kreiman, Gabriel
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2406.16935
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866929398621405184
author Madan, Spandan
Xiao, Will
Cao, Mingran
Pfister, Hanspeter
Livingstone, Margaret
Kreiman, Gabriel
author_facet Madan, Spandan
Xiao, Will
Cao, Mingran
Pfister, Hanspeter
Livingstone, Margaret
Kreiman, Gabriel
contents We characterized the generalization capabilities of DNN-based encoding models when predicting neuronal responses from the visual cortex. We collected \textit{MacaqueITBench}, a large-scale dataset of neural population responses from the macaque inferior temporal (IT) cortex to over $300,000$ images, comprising $8,233$ unique natural images presented to seven monkeys over $109$ sessions. Using \textit{MacaqueITBench}, we investigated the impact of distribution shifts on models predicting neural activity by dividing the images into Out-Of-Distribution (OOD) train and test splits. The OOD splits included several different image-computable types including image contrast, hue, intensity, temperature, and saturation. Compared to the performance on in-distribution test images -- the conventional way these models have been evaluated -- models performed worse at predicting neuronal responses to out-of-distribution images, retaining as little as $20\%$ of the performance on in-distribution test images. The generalization performance under OOD shifts can be well accounted by a simple image similarity metric -- the cosine distance between image representations extracted from a pre-trained object recognition model is a strong predictor of neural predictivity under different distribution shifts. The dataset of images, neuronal firing rate recordings, and computational benchmarks are hosted publicly at: https://bit.ly/3zeutVd.
format Preprint
id arxiv_https___arxiv_org_abs_2406_16935
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Benchmarking Out-of-Distribution Generalization Capabilities of DNN-based Encoding Models for the Ventral Visual Cortex
Madan, Spandan
Xiao, Will
Cao, Mingran
Pfister, Hanspeter
Livingstone, Margaret
Kreiman, Gabriel
Signal Processing
Artificial Intelligence
We characterized the generalization capabilities of DNN-based encoding models when predicting neuronal responses from the visual cortex. We collected \textit{MacaqueITBench}, a large-scale dataset of neural population responses from the macaque inferior temporal (IT) cortex to over $300,000$ images, comprising $8,233$ unique natural images presented to seven monkeys over $109$ sessions. Using \textit{MacaqueITBench}, we investigated the impact of distribution shifts on models predicting neural activity by dividing the images into Out-Of-Distribution (OOD) train and test splits. The OOD splits included several different image-computable types including image contrast, hue, intensity, temperature, and saturation. Compared to the performance on in-distribution test images -- the conventional way these models have been evaluated -- models performed worse at predicting neuronal responses to out-of-distribution images, retaining as little as $20\%$ of the performance on in-distribution test images. The generalization performance under OOD shifts can be well accounted by a simple image similarity metric -- the cosine distance between image representations extracted from a pre-trained object recognition model is a strong predictor of neural predictivity under different distribution shifts. The dataset of images, neuronal firing rate recordings, and computational benchmarks are hosted publicly at: https://bit.ly/3zeutVd.
title Benchmarking Out-of-Distribution Generalization Capabilities of DNN-based Encoding Models for the Ventral Visual Cortex
topic Signal Processing
Artificial Intelligence
url https://arxiv.org/abs/2406.16935