Graduation Year
2023
Document Type
Campus Only Senior Thesis
Degree Name
Bachelor of Arts
Department
Physics
Reader 1
Sarah Marzen
Reader 2
Adam Landsberg
Abstract
The question of whether biological systems are poised at criticality has been a topic of investigation in recent decades. There is evidence to suggest that many biological systems, such as neural networks, amino acids, and flocks of birds operate around the critical point. Criticality is a phase transition point between order and chaos where information processing, in- formation capacity. There is also evidence to suggest that computation is maximized at criticality. In this thesis, we use heat capacity to show that artificial neural networks designed to classify MNIST digits of handwritten numbers are poised at criticality.
Recommended Citation
McVay, Carson, "Artificial neural networks trained to classify MNIST digits are poised at criticality" (2023). Scripps Senior Theses. 2098.
https://scholarship.claremont.edu/scripps_theses/2098
This thesis is restricted to the Claremont Colleges current faculty, students, and staff.