Algorithm Selection with Zero Domain Knowledge via Text Embeddings

TL;DR

This paper introduces ZeroFolio, a domain-agnostic algorithm selection method using pretrained text embeddings, outperforming traditional feature-based approaches across multiple problem domains.

Contribution

It presents a feature-free, text embedding-based pipeline for algorithm selection that works across diverse domains without domain-specific training.

Findings

Outperforms random forest with hand-crafted features in 10 of 11 scenarios

Achieves perfect performance with two-seed voting in all scenarios

Key design choices include inverse-distance weighting and Manhattan distance

Abstract

We propose a feature-free approach to algorithm selection that replaces hand-crafted instance features with pretrained text embeddings. Our method, ZeroFolio, proceeds in three steps: it reads the raw instance file as plain text, embeds it with a pretrained embedding model, and selects an algorithm via weighted k-nearest neighbors. The key to our approach is the observation that pretrained embeddings produce representations that distinguish problem instances without any domain knowledge or task-specific training. This allows us to apply the same three-step pipeline (serialize, embed, select) across diverse problem domains with text-based instance formats. We evaluate our approach on 11 ASlib scenarios spanning 7 domains (SAT, MaxSAT, QBF, ASP, CSP, MIP, and graph problems). Our experiments show that this approach outperforms a random forest trained on hand-crafted features in 10 of 11 scenarios with a single fixed configuration, and in all 11 with two-seed voting; the margin is often substantial. Our ablation study shows that inverse-distance weighting, line shuffling, and Manhattan distance are the key design choices. On scenarios where both selectors are competitive, combining embeddings with hand-crafted features via soft voting yields further improvements.