Skip to content
Back to skills

Attribution Neuron Utility

ASecurity

Gradient-based eXplained Deviation (GXD) neuron utility metric for plasticity restoration in deep networks. Combines attribution-guided neuron importance scoring with critical bottleneck period (CBP) detection and targeted reset. Use when: catastrophic forgetting, plasticity loss, neuron-level analysis, continual learning reset methods, attribution-guided training.

  • 3 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 11, 2026
tools

Security analysis

A100/100

Scanned September 11, 2026

npx -y skills add hiyenwong/ai_collection --skill attribution-neuron-utility --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Attribution Neuron Utility?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Attribution Neuron Utility
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/hiyenwong-attribution-neuron-utility/badge)](https://www.skillsdirectory.com/skills/hiyenwong-attribution-neuron-utility)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: attribution-neuron-utility
description: "Gradient-based eXplained Deviation (GXD) neuron utility metric for plasticity restoration in deep networks. Combines attribution-guided neuron importance scoring with critical bottleneck period (CBP) detection and targeted reset. Use when: catastrophic forgetting, plasticity loss, neuron-level analysis, continual learning reset methods, attribution-guided training."
---

# Attribution-Based Neuron Utility for Plasticity Restoration

## Core Method

**GXD (Gradient-based eXplained Deviation)**: Attribution-based neuron utility metric that identifies which neurons contribute to prediction vs. which are dead.

## CBP Detection

Critical Bottleneck Periods (CBPs) are phases where gradient norms systematically decrease, indicating plasticity loss.

## Reset Strategy

1. Compute GXD scores per neuron
2. Identify low-utility (dead) neurons during CBPs
3. Selective reset of dead neurons while preserving functional ones
4. Outperforms L2 regularization, dropout, and L3 methods

## Paper

- Elisii et al., arXiv:2605.06834, 2026

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in tools →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.