GeomCA: Geometric Evaluation of Data Representations

05/26/2021
by   Petra Poklukar, et al.
0

Evaluating the quality of learned representations without relying on a downstream task remains one of the challenges in representation learning. In this work, we present Geometric Component Analysis (GeomCA) algorithm that evaluates representation spaces based on their geometric and topological properties. GeomCA can be applied to representations of any dimension, independently of the model that generated them. We demonstrate its applicability by analyzing representations obtained from a variety of scenarios, such as contrastive learning models, generative models and supervised learning models.

READ FULL TEXT
POST COMMENT

Comments

There are no comments yet.

Authors

page 1

page 9

03/07/2017

On the Limits of Learning Representations with Label-Based Supervision

Advances in neural network based classifiers have transformed automatic ...
12/16/2021

High Fidelity Visualization of What Your Self-Supervised Representation Knows About

Discovering what is learned by neural networks remains a challenge. In s...
06/19/2019

Unsupervised State Representation Learning in Atari

State representation learning, or the ability to capture latent generati...
02/17/2021

Contrastive Learning Inverts the Data Generating Process

Contrastive learning has recently seen tremendous success in self-superv...
11/15/2016

Interpreting the Syntactic and Social Elements of the Tweet Representations via Elementary Property Prediction Tasks

Research in social media analysis is experiencing a recent surge with a ...
09/18/2021

Interest-oriented Universal User Representation via Contrastive Learning

User representation is essential for providing high-quality commercial s...
06/05/2020

Evaluating the Disentanglement of Deep Generative Models through Manifold Topology

Learning disentangled representations is regarded as a fundamental task ...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.