Information-theoretical label embeddings for large-scale image classification

07/19/2016
by   Francois Chollet, et al.
0

We present a method for training multi-label, massively multi-class image classification models, that is faster and more accurate than supervision via a sigmoid cross-entropy loss (logistic regression). Our method consists in embedding high-dimensional sparse labels onto a lower-dimensional dense sphere of unit-normed vectors, and treating the classification problem as a cosine proximity regression problem on this sphere. We test our method on a dataset of 300 million high-resolution images with 17,000 labels, where it yields considerably faster convergence, as well as a 7 compared to logistic regression.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
11/20/2018

Single-Label Multi-Class Image Classification by Deep Logistic Regression

The objective learning formulation is essential for the success of convo...
research
07/20/2021

Understanding Gender and Racial Disparities in Image Recognition Models

Large scale image classification models trained on top of popular datase...
research
02/14/2019

MultiGrain: a unified image embedding for classes and instances

MultiGrain is a network architecture producing compact vector representa...
research
09/08/2017

Extreme Sparse Multinomial Logistic Regression: A Fast and Robust Framework for Hyperspectral Image Classification

Although the sparse multinomial logistic regression (SMLR) has provided ...
research
09/17/2018

Revisit Multinomial Logistic Regression in Deep Learning: Data Dependent Model Initialization for Image Recognition

We study in this paper how to initialize the parameters of multinomial l...
research
05/29/2022

A Conditional Randomization Test for Sparse Logistic Regression in High-Dimension

Identifying the relevant variables for a classification model with corre...
research
04/17/2019

Correlated Logistic Model With Elastic Net Regularization for Multilabel Image Classification

In this paper, we present correlated logistic (CorrLog) model for multil...

Please sign up or login with your details

Forgot password? Click here to reset