Unsupervised Embedding Learning via Invariant and Spreading Instance Feature

TL;DR

This paper introduces a novel unsupervised embedding learning method that optimizes instance features using a softmax approach, achieving faster training and higher accuracy for both seen and unseen categories.

Contribution

It proposes a new instance-based softmax embedding method that directly optimizes real instance features, improving learning speed and accuracy over existing methods.

Findings

Faster learning speed compared to existing methods

Higher accuracy in embedding tasks

Effective on both seen and unseen categories

Abstract

This paper studies the unsupervised embedding learning problem, which requires an effective similarity measurement between samples in low-dimensional embedding space. Motivated by the positive concentrated and negative separated properties observed from category-wise supervised learning, we propose to utilize the instance-wise supervision to approximate these properties, which aims at learning data augmentation invariant and instance spread-out features. To achieve this goal, we propose a novel instance based softmax embedding method, which directly optimizes the `real' instance features on top of the softmax function. It achieves significantly faster learning speed and higher accuracy than all existing methods. The proposed method performs well for both seen and unseen testing categories with cosine similarity. It also achieves competitive performance even without pre-trained network over samples from fine-grained categories.