Mahdi Hadj Ali, Yann Le Biannic, Pierre-Henri Wuillemin
{"title":"利用预测模型可靠地估计因果效应","authors":"Mahdi Hadj Ali, Yann Le Biannic, Pierre-Henri Wuillemin","doi":"10.1142/s0218213024600066","DOIUrl":null,"url":null,"abstract":"In recent years, machine learning algorithms have been widely adopted across many fields due to their efficiency and versatility. However, the complexity of predictive models has led to a lack of interpretability in automatic decision-making. Recent works have improved general interpretability by estimating the contributions of input features to the predictions of a pre-trained model. Drawing on these improvements, practitioners seek to gain causal insights into the underlying data-generating mechanisms. To this end, works have attempted to integrate causal knowledge into interpretability, as non-causal techniques can lead to paradoxical explanations. In this paper, we argue that each question about a causal effect requires its own reasoning and that relying on an initial predictive model trained on an arbitrary set of variables may result in quantification problems when estimating all possible effects. As an alternative, we advocate for a query-driven methodology that addresses each causal question separately. Assuming that the causal structure relating the variables is known, we propose to employ the tools of causal inference to quantify a particular effect as a formula involving observable probabilities. We then derive conditions on the selection of variables to train a predictive model that is tailored for the causal question of interest. Finally, we identify suitable eXplainable AI (XAI) techniques to estimate causal effects from the model predictions. Furthermore, we introduce a novel method for estimating direct effects through intervention on causal mechanisms.","PeriodicalId":50280,"journal":{"name":"International Journal on Artificial Intelligence Tools","volume":null,"pages":null},"PeriodicalIF":1.0000,"publicationDate":"2024-04-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Reliable Estimation of Causal Effects Using Predictive Models\",\"authors\":\"Mahdi Hadj Ali, Yann Le Biannic, Pierre-Henri Wuillemin\",\"doi\":\"10.1142/s0218213024600066\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In recent years, machine learning algorithms have been widely adopted across many fields due to their efficiency and versatility. However, the complexity of predictive models has led to a lack of interpretability in automatic decision-making. Recent works have improved general interpretability by estimating the contributions of input features to the predictions of a pre-trained model. Drawing on these improvements, practitioners seek to gain causal insights into the underlying data-generating mechanisms. To this end, works have attempted to integrate causal knowledge into interpretability, as non-causal techniques can lead to paradoxical explanations. In this paper, we argue that each question about a causal effect requires its own reasoning and that relying on an initial predictive model trained on an arbitrary set of variables may result in quantification problems when estimating all possible effects. As an alternative, we advocate for a query-driven methodology that addresses each causal question separately. Assuming that the causal structure relating the variables is known, we propose to employ the tools of causal inference to quantify a particular effect as a formula involving observable probabilities. We then derive conditions on the selection of variables to train a predictive model that is tailored for the causal question of interest. Finally, we identify suitable eXplainable AI (XAI) techniques to estimate causal effects from the model predictions. Furthermore, we introduce a novel method for estimating direct effects through intervention on causal mechanisms.\",\"PeriodicalId\":50280,\"journal\":{\"name\":\"International Journal on Artificial Intelligence Tools\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.0000,\"publicationDate\":\"2024-04-25\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"International Journal on Artificial Intelligence Tools\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://doi.org/10.1142/s0218213024600066\",\"RegionNum\":4,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal on Artificial Intelligence Tools","FirstCategoryId":"94","ListUrlMain":"https://doi.org/10.1142/s0218213024600066","RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}
Reliable Estimation of Causal Effects Using Predictive Models
In recent years, machine learning algorithms have been widely adopted across many fields due to their efficiency and versatility. However, the complexity of predictive models has led to a lack of interpretability in automatic decision-making. Recent works have improved general interpretability by estimating the contributions of input features to the predictions of a pre-trained model. Drawing on these improvements, practitioners seek to gain causal insights into the underlying data-generating mechanisms. To this end, works have attempted to integrate causal knowledge into interpretability, as non-causal techniques can lead to paradoxical explanations. In this paper, we argue that each question about a causal effect requires its own reasoning and that relying on an initial predictive model trained on an arbitrary set of variables may result in quantification problems when estimating all possible effects. As an alternative, we advocate for a query-driven methodology that addresses each causal question separately. Assuming that the causal structure relating the variables is known, we propose to employ the tools of causal inference to quantify a particular effect as a formula involving observable probabilities. We then derive conditions on the selection of variables to train a predictive model that is tailored for the causal question of interest. Finally, we identify suitable eXplainable AI (XAI) techniques to estimate causal effects from the model predictions. Furthermore, we introduce a novel method for estimating direct effects through intervention on causal mechanisms.
期刊介绍:
The International Journal on Artificial Intelligence Tools (IJAIT) provides an interdisciplinary forum in which AI scientists and professionals can share their research results and report new advances on AI tools or tools that use AI. Tools refer to architectures, languages or algorithms, which constitute the means connecting theory with applications. So, IJAIT is a medium for promoting general and/or special purpose tools, which are very important for the evolution of science and manipulation of knowledge. IJAIT can also be used as a test ground for new AI tools.
Topics covered by IJAIT include but are not limited to: AI in Bioinformatics, AI for Service Engineering, AI for Software Engineering, AI for Ubiquitous Computing, AI for Web Intelligence Applications, AI Parallel Processing Tools (hardware/software), AI Programming Languages, AI Tools for CAD and VLSI Analysis/Design/Testing, AI Tools for Computer Vision and Speech Understanding, AI Tools for Multimedia, Cognitive Informatics, Data Mining and Machine Learning Tools, Heuristic and AI Planning Strategies and Tools, Image Understanding, Integrated/Hybrid AI Approaches, Intelligent System Architectures, Knowledge-Based/Expert Systems, Knowledge Management and Processing Tools, Knowledge Representation Languages, Natural Language Understanding, Neural Networks for AI, Object-Oriented Programming for AI, Reasoning and Evolution of Knowledge Bases, Self-Healing and Autonomous Systems, and Software Engineering for AI.