Human Assisted Learning by Evolutionary Multi-Objective Optimization
نویسندگان
چکیده
Machine learning models have liberated manpower greatly in many real-world tasks, but their predictions are still worse than humans on some specific instances. To improve the performance, it is natural to optimize machine take decisions for most instances while delivering a few tricky humans, resulting problem of Human Assisted Learning (HAL). Previous works mainly formulated HAL as constrained optimization that tries find limited subset human decision such sum model and errors can be minimized; employed greedy algorithms, whose however, may due nature. In this paper, we propose new framework HAL-EMO based Evolutionary Multi-objective Optimization, which reformulates bi-objective minimizes number selected total simultaneously, employs Multi-Objective Algorithm (MOEA) solve it. We implement using two MOEAs, popular NSGA-II well theoretically grounded GSEMO. also MOEA, called BSEMO, with biased selection balanced mutation HAL-EMO, prove assisted regression classification, BSEMO achieve better same theoretical guarantees previous respectively. Experiments tasks medical diagnosis content moderation show superiority (with either NSGA-II, GSEMO or BSEMO) over leads best performance HAL-EMO.
منابع مشابه
solution of security constrained unit commitment problem by a new multi-objective optimization method
چکیده-پخش بار بهینه به عنوان یکی از ابزار زیر بنایی برای تحلیل سیستم های قدرت پیچیده ،برای مدت طولانی مورد بررسی قرار گرفته است.پخش بار بهینه توابع هدف یک سیستم قدرت از جمله تابع هزینه سوخت ،آلودگی ،تلفات را بهینه می کند،و هم زمان قیود سیستم قدرت را نیز برآورده می کند.در کلی ترین حالتopf یک مساله بهینه سازی غیر خطی ،غیر محدب،مقیاس بزرگ،و ایستا می باشد که می تواند شامل متغیرهای کنترلی پیوسته و گ...
Single- and Multi-objective Evolutionary Design Optimization Assisted by Gaussian Random Field Metamodels
متن کامل
Multi-objective Optimization of web profile of railway wheel using Bi-directional Evolutionary Structural Optimization
In this paper, multi-objective optimization of railway wheel web profile using bidirectional evolutionary structural optimization (BESO) algorithm is investigated. Using a finite element software, static analysis of the wheel based on a standard load case, and its modal analysis for finding the fundamental natural frequency is performed. The von Mises stress and critical frequency as the proble...
متن کاملMulti-objective Evolutionary Optimization of Subsonic Airfoils
This work concerns the application of multi-objective evolutionary optimization by approximation function to 2D aerodynamic design. The new general concept of evolution control is used to on-line enriching the database of correct solutions, which are the basis of the learning procedure for the approximators. Substantially, given an initial very poor model approximation, which means small size o...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2023
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v37i10.26467