REVIEW 5 cited by
Dying ReLU and Initialization: Theory and Numerical Examples
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
The dying ReLU refers to the problem when ReLU neurons become inactive and only output 0 for any input. There are many empirical and heuristic explanations of why ReLU neurons die. However, little is known about its theoretical analysis. In this paper, we rigorously prove that a deep ReLU network will eventually die in probability as the depth goes to infinite. Several methods have been proposed to alleviate the dying ReLU. Perhaps, one of the simplest treatments is to modify the initialization procedure. One common way of initializing weights and biases uses symmetric probability distributions, which suffers from the dying ReLU. We thus propose a new initialization procedure, namely, a randomized asymmetric initialization. We prove that the new initialization can effectively prevent the dying ReLU. All parameters required for the new initialization are theoretically designed. Numerical examples are provided to demonstrate the effectiveness of the new initialization procedure.
Forward citations
Cited by 5 Pith papers
-
The Map Behind the Flow: Finite-Step Gradient Descent as a Dynamical System
The edge of stability is the first bifurcation of the finite-step gradient map; residual oscillations then drive balancing and representation selection beyond that edge.
-
Optimized Weight Initialization on the Stiefel Manifold for Deep ReLU Neural Networks
An initialization that maximizes a semi-orthogonal weight matrix's alignment with the all-ones vector prevents dying ReLU and keeps 100-layer ReLU networks trainable.
-
STAR-Pose: Efficient Low-Resolution Video Human Pose Estimation via Spatial-Temporal Adaptive Super-Resolution
STAR-Pose combines a spatial-temporal transformer with linear attention, a CNN texture branch, and a pose-aware loss to estimate human pose from low-resolution video, reporting up to 5.2% mAP improvement at 64x48.
-
4D-MISR: A unified model for low-dose super-resolution imaging via feature fusion
4D-MISR maps low-dose, multi-view virtual bright-field 4D-STEM data to high-resolution images using a CNN trained on simulated specimens, but the evaluation may be compromised by missing train/test separation.
-
TinyProto: Communication-Efficient Federated Learning with Sparse Prototypes in Resource-Constrained Environments
Compressing class prototypes with per-class masks and a sample-count scaling trick cuts communication cost in prototype-based federated learning by up to several times without hurting accuracy.
Discussion (0). Sign in to comment.