Incremental Embedding Learning Via Zero-shot Translation

Abstract

Modern deep learning methods have achieved great success in machine learning and computer vision fields by learning a set of pre-defined datasets. Howerver, these methods perform unsatisfactorily when applied into real-world situations. The reason of this phenomenon is that learning new tasks leads the trained model quickly forget the knowledge of old tasks, which is referred to as catastrophic forgetting. Current state-of-the-art incremental learning methods tackle catastrophic forgetting problem in traditional classification networks and ignore the problem existing in embedding networks, which are the basic networks for image retrieval, face recognition, zero-shot learning, etc. Different from traditional incremental classification networks, the semantic gap between the embedding spaces of two adjacent tasks is the main challenge for embedding networks under incremental learning setting. Thus, we propose a novel class-incremental method for embedding network, named as zero-shot trans

Stats

• citations11
• S2 citations—
• github stars0
• HF likes0
• heat score8.09
• arxiv keywei2020incremental

Related papers

• Visual Space Optimization For Zero-shot Learning (2019)0.00
• Semantic Diversity Learning For Zero-shot Multi-label Classification (2021)11.85
• Hybrid-attention Based Decoupled Metric Learning For Zero-shot Image Retrieval (2019)12.93
• Transductive Zero-shot Hashing Via Coarse-to-fine Similarity Mining (2017)5.84
• Survey Of Visual-semantic Embedding Methods For Zero-shot Image Retrieval (2021)4.52
• Beginning With You: Perceptual-initialization Improves Vision-language Representation And Alignment (2025)0.00
• A Probabilistic Approach For Learning Embeddings Without Supervision (2019)0.00
• Liteembed: Adapting CLIP To Rare Classes (2026)0.00