At matched task accuracy, e-prop trained RNNs reach neural data similarity comparable to BPTT trained RNNs on Mante 2013 and Sussillo 2015 datasets, with initialization and architecture influencing similarity more than the learning rule.
Although the neural datasets used in this study contain multiple trials per condition, we did not compute data-data similarity via trial-based subsampling
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.NE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Can Biologically Plausible Temporal Credit Assignment Rules Match BPTT for Neural Similarity? E-prop as an Example
At matched task accuracy, e-prop trained RNNs reach neural data similarity comparable to BPTT trained RNNs on Mante 2013 and Sussillo 2015 datasets, with initialization and architecture influencing similarity more than the learning rule.