Evolutionary Retrofitting

TL;DR

AfterLearnER introduces an evolutionary optimization method to refine trained machine learning models using non-differentiable error signals, applicable post-training or during inference, with minimal feedback requirements and broad versatility.

Contribution

It presents a novel evolutionary retrofitting approach that enhances models with non-differentiable feedback, supporting dynamic, post-training, and inference-time optimization.

Findings

Effective on non-differentiable signals like threshold criteria and BLEU scores

Requires only a small amount of feedback, from dozens to hundreds of scalars

Demonstrates versatility across various applications and models

Abstract

AfterLearnER (After Learning Evolutionary Retrofitting) consists in applying evolutionary optimization to refine fully trained machine learning models by optimizing a set of carefully chosen parameters or hyperparameters of the model, with respect to some actual, exact, and hence possibly non-differentiable error signal, performed on a subset of the standard validation set. The efficiency of AfterLearnER is demonstrated by tackling non-differentiable signals such as threshold-based criteria in depth sensing, the word error rate in speech re-synthesis, the number of kills per life at Doom, computational accuracy or BLEU in code translation, image quality in 3D generative adversarial networks (GANs), and user feedback in image generation via Latent Diffusion Models (LDM). This retrofitting can be done after training, or dynamically at inference time by taking into account the user feedback. The advantages of AfterLearnER are its versatility, the possibility to use non-differentiable feedback, including human evaluations (i.e., no gradient is needed), the limited overfitting supported by a theoretical study, and its anytime behavior. Last but not least, AfterLearnER requires only a small amount of feedback, i.e., a few dozen to a few hundred scalars, compared to the tens of thousands needed in most related published works.