analiza aiqbench
The paper proposes treating iterative optimization algorithms as trainable autoregressive models that learn from execution trajectories, combining imitation of several hand-designed teachers with later refinement against optimization feedback. The abstract claims gains beyond the teacher algorithms and generalization to unseen, substantially higher dimensions, but these rest on experiments whose benchmarks, baselines, and compute costs aren't described here. Readers working on learned optimizers or meta-learning should look for those details before judging how far the approach extends.
Algorithm Modelization: Transforming Optimization Algorithms into Evolvable Intelligence
arXiv:2610.23161v2 Announce Type: replace Abstract: Can algorithms be learned and evolved as models, rather than repeatedly redesigned as programs? We explore this question by introducing Algorithm Modelization, a new paradigm that turns iterative algorithms into trainable models. Its central idea is to regard algorithm execution trajectories as computational experience: by learning how algorithmic states evolve over iterations, a model can internalize the knowledge embodied in existing algorithms, reproduce their computational behaviors, and further evolve beyond them through training. As a first realization of this paradigm, we modelize iterative optimization as an autoregressive process, where preceding generations constitute the computational context for generating the next population. We develop a dimension-independent population-temporal architecture to capture both population interactions and iterative dynamics, together with an imitation-to-surpassing learning framework that first acquires algorithmic knowledge from heterogeneous teachers and then improves the resulting optimizer directly through optimization feedback. The modelized optimizer consequently goes beyond reproducing individual algorithms: it consolidates heterogeneous search knowledge into a single trainable model, achieves stronger optimization capability through subsequent improvement, and generalizes its acquired knowledge across problems and substantially larger unseen dimensions. More fundamentally, Algorithm Modelization makes algorithmic capability accumulable through learning, where new algorithms and problem-solving experience can become training resources for progressively improving the model. Our results establish a promising path from manually designed algorithms toward general algorithm models that learn, generalize, and evolve from computational experience.
Abstract:Can algorithms be learned and evolved as models, rather than repeatedly redesigned as programs? We explore this question by introducing Algorithm Modelization, a new paradigm that turns iterative algorithms into trainable models. Its central idea is to regard algorithm execution trajectories as computational experience: by learning how algorithmic states evolve over iterations, a model can internalize the knowledge embodied in existing algorithms, reproduce their computational behaviors, and further evolve beyond them through training. As a first realization of this paradigm, we modelize iterative optimization as an autoregressive process, where preceding generations constitute the computational context for generating the next population. We develop a dimension-independent population-temporal architecture to capture both population interactions and iterative dynamics, together with an imitation-to-surpassing learning framework that first acquires algorithmic knowledge from heterogeneous teachers and then improves the resulting optimizer directly through optimization feedback. The modelized optimizer consequently goes beyond reproducing individual algorithms: it consolidates heterogeneous search knowledge into a single trainable model, achieves stronger optimization capability through subsequent improvement, and generalizes its acquired knowledge across problems and substantially larger unseen dimensions. More fundamentally, Algorithm Modelization makes algorithmic capability accumulable through learning, where new algorithms and problem-solving experience can become training resources for progressively improving the model. Our results establish a promising path from manually designed algorithms toward general algorithm models that learn, generalize, and evolve from computational experience.
Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
arXiv:2610.23161 [cs.AI]
(or arXiv:2610.23161v2 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.23161 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 320 cuvinte
Subiecte conexe
Native-speed vLLM transformers modeling backend
Hugging Face · 8 iul. 2026
Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers
Hugging Face · 18 aug. 2026
All-in-One: A Composable Attack on LLM Transformations
arXiv · 27 oct. 2026
Can we change an AI model’s architecture without retraining it from scratch?
Amazon Science · 12 oct. 2026