SymbolNet: Bridging Latent Neural Representations and Symbolic Reasoning via Intermediate Feature Interpretation

SymbolNet: Bridging Latent Neural Representations and Symbolic Reasoning via Intermediate Feature Interpretation

The interpretation of intermediate representations in deep neural networks is critical for enhancing the transparency, trustworthiness, and applicability of artificial intelligence (AI) systems. In this paper, we propose SymbolNet, a framework that extracts mid-level features from trained models and...

Full description

Saved in:
Bibliographic Details
Main Authors: Sungheon Jeong, Hyungjoon Kim, Yoojeong Song
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Computer vision
symbolic representation
explainable AI
intermediate feature extraction
neural network interpretability
Online Access:https://ieeexplore.ieee.org/document/10980088/
Tags: Add Tag
No Tags, Be the first to tag this record!

Similar Items

Similar Items