Naive Few-Shot Learning: Uncovering the fluid intelligence of machines

TL;DR

This paper demonstrates that naive, randomly initialized deep neural networks can solve sequence regularity tasks and real-world anomaly detection without prior training, highlighting their potential for fluid intelligence.

Contribution

It introduces a benchmark for sequence regularity detection and shows that untrained deep networks can perform well on these tasks, revealing their innate fluid intelligence capabilities.

Findings

Naive deep networks can solve sequence regularity tasks after a single training step.

Untrained networks can detect anomalies in visual and auditory data.

Results suggest innate fluid intelligence in deep neural networks.

Abstract

In this paper, we aimed to help bridge the gap between human fluid intelligence - the ability to solve novel tasks without prior training - and the performance of deep neural networks, which typically require extensive prior training. An essential cognitive component for solving intelligence tests, which in humans are used to measure fluid intelligence, is the ability to identify regularities in sequences. This motivated us to construct a benchmark task, which we term \textit{sequence consistency evaluation} (SCE), whose solution requires the ability to identify regularities in sequences. Given the proven capabilities of deep networks, their ability to solve such tasks after extensive training is expected. Surprisingly, however, we show that naive (randomly initialized) deep learning models that are trained on a \textit{single} SCE with a \textit{single} optimization step can still solve non-trivial versions of the task relatively well. We extend our findings to solve, without any prior training, real-world anomaly detection tasks in the visual and auditory modalities. These results demonstrate the fluid-intelligent computational capabilities of deep networks. We discuss the implications of our work for constructing fluid-intelligent machines.