2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2019
DOI: 10.1109/cvpr.2019.00555
|View full text |Cite
|
Sign up to set email alerts
|

A Neurobiological Evaluation Metric for Neural Network Model Search

Abstract: Neuroscience theory posits that the brain's visual system coarsely identifies broad object categories via neural activation patterns, with similar objects producing similar neural responses. Artificial neural networks also have internal activation behavior in response to stimuli. We hypothesize that networks exhibiting brain-like activation behavior will demonstrate brain-like characteristics, e.g., stronger

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
5
0

Year Published

2021
2021
2023
2023

Publication Types

Select...
7
2

Relationship

2
7

Authors

Journals

citations
Cited by 10 publications
(5 citation statements)
references
References 54 publications
(103 reference statements)
0
5
0
Order By: Relevance
“…Our work also provide valuable insights into how one could consider the influence of architecture on a learned representation. If high performing models have ultimately learned similar representations, areas like neural architecture search (NAS) may consider shifting their focus to identifying higher performing representations-there is some work in this domain, such as the hypothesis by Blanchard et al (2019) that learned representations that mirror biology, by grouping similar-looking objects in the . /fcomp.…”
Section: Discussionmentioning
confidence: 99%
“…Our work also provide valuable insights into how one could consider the influence of architecture on a learned representation. If high performing models have ultimately learned similar representations, areas like neural architecture search (NAS) may consider shifting their focus to identifying higher performing representations-there is some work in this domain, such as the hypothesis by Blanchard et al (2019) that learned representations that mirror biology, by grouping similar-looking objects in the . /fcomp.…”
Section: Discussionmentioning
confidence: 99%
“…An AI model is "biologically inspired" if its design, training, or conguration is based on observations of the living world [11,19]. Bio-inspired models form the cutting edge in several areas of AI research, for example by modifying the training process for object recognition to be more consistent with data from fMRI scans of humans performing the same recognition [10,20,41,57]. Promising evidence exists comparing human programmer and machine attention [42], albeit using a survey methodology instead of eye-tracking data as in this paper.…”
Section: Biologically-inspired Learningmentioning
confidence: 99%
“…With the development and progress of computer technology, the field of computer image visual analysis has developed rapidly, which is a frontier scientific and technological field with a wide range of applications. The relevant knowledge of AI technology has been extensively used in this field, and video behaviour analysis technology is an important foundation of artificial intelligence application [5][6][7][8]. The interpretation of video behaviour analysis is to imitate the structure of the human brain through computer image analysis technology, and to examine and analyze the actions of the characters in the video [9][10][11][12].…”
Section: Introductionmentioning
confidence: 99%