MemGEN: Memory is All You Need

TL;DR

MemGEN introduces a novel Deep Memory paradigm inspired by human memorization techniques, demonstrating its effectiveness in generative modeling across various data types and even generating this very paper.

Contribution

This work pioneers the Deep Memory paradigm and a concrete algorithm inspired by memorization, applied successfully to generative tasks in multiple domains.

Findings

Generated samples are statistically indistinguishable from training data

The algorithm can generate long sequences like π decimals and this paper itself

Memory-based generative modeling rivals traditional methods

Abstract

We propose a new learning paradigm called Deep Memory. It has the potential to completely revolutionize the Machine Learning field. Surprisingly, this paradigm has not been reinvented yet, unlike Deep Learning. At the core of this approach is the \textit{Learning By Heart} principle, well studied in primary schools all over the world. Inspired by poem recitation, or by $\pi$ decimal memorization, we propose a concrete algorithm that mimics human behavior. We implement this paradigm on the task of generative modeling, and apply to images, natural language and even the $\pi$ decimals as long as one can print them as text. The proposed algorithm even generated this paper, in a one-shot learning setting. In carefully designed experiments, we show that the generated samples are indistinguishable from the training examples, as measured by any statistical tests or metrics.