Geometric Generalization Based Zero-Shot Learning Dataset Infinite World: Simple Yet Powerful

TL;DR

This paper introduces Infinite World, a scalable geometric dataset and zero-shot learning tests inspired by Raven's Matrices, to evaluate deep models' ability to generalize concepts and reason spatially and numerically.

Contribution

It presents a novel, scalable geometric dataset and a zero-shot intelligence metric to assess and analyze the generalization and reasoning capabilities of deep generative models.

Findings

State-of-the-art models show limited generalization in geometric reasoning.

The dataset scales to infinity in features and size.

Proposed optimization improves few-shot and zero-shot learning performance.

Abstract

Raven's Progressive Matrices are one of the widely used tests in evaluating the human test taker's fluid intelligence. Analogously, this paper introduces geometric generalization based zero-shot learning tests to measure the rapid learning ability and the internal consistency of deep generative models. Our empirical research analysis on state-of-the-art generative models discern their ability to generalize concepts across classes. In the process, we introduce Infinite World, an evaluable, scalable, multi-modal, light-weight dataset and Zero-Shot Intelligence Metric ZSI. The proposed tests condenses human-level spatial and numerical reasoning tasks to its simplistic geometric forms. The dataset is scalable to a theoretical limit of infinity, in numerical features of the generated geometric figures, image size and in quantity. We systematically analyze state-of-the-art model's internal consistency, identify their bottlenecks and propose a pro-active optimization method for few-shot and zero-shot learning.