Concept-Oriented Deep Learning

2018/06/05 by Daniel T. Chang, Chang, Daniel T
Computer Science · #Advanced Graph Neural Networks #Artificial Intelligence (cs.AI) #Bayesian Modeling and Causal Inference #FOS: Computer and information sciences #Topic Modeling

paper · pdf · doi:10.48550/arxiv.1806.01756

openalex publication_date 2018/06/05 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Concepts are the foundation of human deep learning, understanding, and knowledge integration and transfer. We propose concept-oriented deep learning (CODL) which extends (machine) deep learning with concept representations and conceptual understanding capability. CODL addresses some of the major limitations of deep learning: interpretability, transferability, contextual adaptation, and requirement for lots of labeled training data. We discuss the major aspects of CODL including concept graph, concept representations, concept exemplars, and concept representation learning systems supporting incremental and continual learning.

Citations

• An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks
• Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks
• The Mythos of Model Interpretability
• A Review on Deep Learning Techniques Applied to Semantic Segmentation

Related

• Concept-Oriented Deep Learning with Large Language Models
• Concept Representation Learning with Contrastive Self-Supervised Learning
• Concept-Oriented Deep Learning: Generative Concept Representations
• Simple Mechanisms for Representing, Indexing and Manipulating Concepts
• Dual Embodied-Symbolic Concept Representations for Deep Learning