ADCP extracts single-neuron-per-layer critical paths via genetic mutation and uses an SVDD voting ensemble to separate normal from adversarial, out-of-distribution, and noisy inputs.
Distilling Critical Paths in Convolutional Neural Networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Neural network compression and acceleration are widely demanded currently due to the resource constraints on most deployment targets. In this paper, through analyzing the filter activation, gradients, and visualizing the filters' functionality in convolutional neural networks, we show that the filters in higher layers learn extremely task-specific features, which are exclusive for only a small subset of the overall tasks, or even a single class. Based on such findings, we reveal the critical paths of information flow for different classes. And by their intrinsic property of exclusiveness, we propose a critical path distillation method, which can effectively customize the convolutional neural networks to small ones with much smaller model size and less computation.
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Anomaly Detection Based on Critical Paths for Deep Neural Networks
ADCP extracts single-neuron-per-layer critical paths via genetic mutation and uses an SVDD voting ensemble to separate normal from adversarial, out-of-distribution, and noisy inputs.