Constrained Few-Shot Learning: Human-Like Low Sample Complexity Learning and Non-Episodic Text Classification

TL;DR

This paper introduces constrained few-shot learning (CFSL), a new paradigm that aligns more closely with human learning by limiting training instances, and proposes a Cat2Vec-based method with a novel contrastive loss inspired by cognitive theories.

Contribution

It defines CFSL as a new task setting with constrained training data and develops a Cat2Vec method utilizing a novel contrastive loss inspired by cognitive theories.

Findings

CFSL better mimics human learning processes.

The proposed Cat2Vec method improves low-sample classification.

Contrastive loss based on cognitive theories enhances learning.

Abstract

Few-shot learning (FSL) is an emergent paradigm of learning that attempts to learn to reason with low sample complexity to mimic the way humans learn, generalise and extrapolate from only a few seen examples. While FSL attempts to mimic these human characteristics, fundamentally, the task of FSL as conventionally formulated using meta-learning with episodic-based training does not in actuality align with how humans acquire and reason with knowledge. FSL with episodic training, while only requires $K$ instances of each test class, still requires a large number of labelled training instances from disjoint classes. In this paper, we introduce the novel task of constrained few-shot learning (CFSL), a special case of FSL where $M$, the number of instances of each training class is constrained such that $M \leq K$ thus applying a similar restriction during FSL training and test. We propose a method for CFSL leveraging Cat2Vec using a novel categorical contrastive loss inspired by cognitive theories such as fuzzy trace theory and prototype theory.