How do Simple Rules `Fit to Reality' in a Complex World?

The theory of fast and frugal heuristics, developed in a new book called Simple Heuristics that make Us Smart (Gigerenzer, Todd, and the ABC Research Group, in press), includes two requirements for rational decision making. One is that decision rules are bounded in their rationality –- that rules are frugal in what they take into account, and therefore fast in their operation. The second is that the rules are ecologically adapted to the environment, which means that they `fit to reality.' The main purpose of this article is to apply these ideas to learning rules–-methods for constructing, selecting, or evaluating competing hypotheses in science, and to the methodology of machine learning, of which connectionist learning is a special case. The bad news is that ecological validity is particularly difficult to implement and difficult to understand. The good news is that it builds an important bridge from normative psychology and machine learning to recent work in the philosophy of science, which considers predictive accuracy to be a primary goal of science.     

Learning,goals, and learning goals: A perspective on goal-driven learning

In cognitive science, artificial intelligence, psychology, and education, a growing body of research supports the view that the learning process is strongly influenced by the learner's goals. The fundamental tenet ofgoal-driven learning is that learning is largely an active and strategic process in which the learner, human or machine, attempts to identify and satisfy its information needs in the context of its tasks and goals, its prior knowledge, its capabilities, and environmental opportunities for learning. This article examines the motivations for adopting a goal-driven model of learning, the relationship between task goals and learning goals, the influences goals can have on learning, and the pragmatic implications of the goal-driven learning model. It presents a new integrative framework for understanding the goal-driven learning process and applies this framework to characterizing research on goal-driven learning.     

Introduction: Machine Learning as Philosophy of Science

I consider three aspects in which machine learning and philosophy of science can illuminate each other: methodology, inductive simplicity and theoretical terms. I examine the relations between the two subjects and conclude by claiming these relations to be very close.     

Semi-supervised learning with an imperfect supervisor

Real-life applications may involve huge data sets with misclassified or partially classified training data. Semi-supervised learning and learning in the presence of label noise have recently emerged as new paradigms in the machine learning community to cope with this kind of problems. This paper describes a new discriminant algorithm for semi-supervised learning. This algorithm optimizes the classification maximum likelihood (CML) of a set of labeled–unlabeled data, using a discriminant extension of the Classification Expectation Maximization algorithm. We further propose to extend this algorithm by modeling imperfections in the estimated class labels for unlabeled data. The parameters of this label-error model are learned together with the semi-supervised classifier parameters. We demonstrate the effectiveness of the approach using extensive experiments on different datasets. Massih R. Amini is currently assistant professor in the University of Pierre and Marie Curie (Paris 6). He received an engineering degree in computer science from the Ecole Supérieure d'Informatique (Computer science engineering school) in Paris in 1995. He then accomplished his master thesis in science in artificial intelligence in 1997 and obtained his PhD in 2001 at University of Pierre and Marie Curie. His research interests include Statistical Learning and Text-Mining. Patrick Gallinari is currently professor in the University of Pierre and Marie Curie (Paris 6) and head of the Computer Science laboratory (LIP6). His main research activity has been in the field of statistical machine learning for the last 15 years. He has also contributed in developing machine learning techniques for different application domains like information retrieval and text mining, user modelling, man–machine interaction and pen interfaces.     

Choosing from Competing Theories in Computerised Learning

In this paper we refer to a machine learning method that reveals all the if–then rules in the data, and on the basis of these rules issues predictions for new cases. When issuing predictions this method faces the problem of choosing from competing theories. We dealt with this problem by calculating the probability that the rule is accidental. The lower this probability, the more the rule can be `trusted' when issuing predictions. The method was tested empirically and found to be accurate. On a broader scope this approach demonstrates how the dialog between researchers in machine learning and the philosophy of science can be beneficial for both sides.     

Data-driven algorithm selection and tuning in optimization and signal processing

Machine learning algorithms typically rely on optimization subroutines and are well known to provide very effective outcomes for many types of problems. Here, we flip the reliance and ask the reverse question: can machine learning algorithms lead to more effective outcomes for optimization problems? Our goal is to train machine learning methods to automatically improve the performance of optimization and signal processing algorithms. As a proof of concept, we use our approach to improve two popular data processing subroutines in data science: stochastic gradient descent and greedy methods in compressed sensing. We provide experimental results that demonstrate the answer is “yes”, machine learning algorithms do lead to more effective outcomes for optimization problems, and show the future potential for this research direction. In addition to our experimental work, we prove relevant Probably Approximately Correct (PAC) learning theorems for our problems of interest. More precisely, we show that there exists a learning algorithm that, with high probability, will select the algorithm that optimizes the average performance on an input set of problem instances with a given distribution.

     

Philosophy of Mind Is (in Part) Philosophy of Computer Science

In this paper I argue that whether or not a computer can be built that passes the Turing test is a central question in the philosophy of mind. Then I show that the possibility of building such a computer depends on open questions in the philosophy of computer science: the physical Church-Turing thesis and the extended Church-Turing thesis. I use the link between the issues identified in philosophy of mind and philosophy of computer science to respond to a prominent argument against the possibility of building a machine that passes the Turing test. Finally, I respond to objections against the proposed link between questions in the philosophy of mind and philosophy of computer science.     

Data-driven approaches to empirical discovery

In this paper we track the development of research in empirical discovery. We focus on four machine discovery systems that share a number of features: the use of data-driven heuristics to constrain the search for numeric laws; a reliance on theoretical terms; and the recursive application of a few general discovery methods. We examine each system in light of the innovations it introduced over its predecessors, providing some insight into the conceptual progress that has occurred in machine discovery. Finally, we reexamine this research from the perspectives of the history and philosophy of science.     

基于k个标记样本的弱监督学习框架

聚类是机器学习领域中的一个研究热点,弱监督学习是半监督学习中一个重要的研究方向,有广泛的应用场景.在对聚类与弱监督学习的研究中,提出了一种基于k个标记样本的弱监督学习框架.该框架首先用聚类及聚类置信度实现了标记样本的扩展.其次,对受限玻尔兹曼机的能量函数进行改进,提出了基于k个标记样本的受限玻尔兹曼机学习模型.最后,完成了对该模型的推理并设计相关算法.为了完成对该框架和模型的检验,选择公开的数据集进行对比实验,实验结果表明,基于k个标记样本的弱监督学习框架实验效果较好.     

The Open International Soccer Database for machine learning

How well can machine learning predict the outcome of a soccer game, given the most commonly and freely available match data? To help answer this question and to facilitate machine learning research in soccer, we have developed the Open International Soccer Database. Version v1.0 of the Database contains essential information from 216,743 league soccer matches from 52 leagues in 35 countries. The earliest entries in the Database are from the year 2000, which is when football leagues generally adopted the “three points for a win” rule. To demonstrate the use of the Database for machine learning research, we organized the 2017 Soccer Prediction Challenge. One of the goals of the Challenge was to estimate where the limits of predictability lie, given the type of match data contained in the Database. Another goal of the Challenge was to pose a real-world machine learning problem with a fixed time line and a genuine prediction task: to develop a predictive model from the Database and then to predict the outcome of the 206 future soccer matches taking place from 31 March 2017 to the end of the regular season. The Open International Soccer Database is released as an open science project, providing a valuable resource for soccer analysts and a unique benchmark for advanced machine learning methods. Here, we describe the Database and the 2017 Soccer Prediction Challenge and its results.

     

Book Review on “Out of Their Minds: The Lives and Discoveries of 15 Great Computer Scientists”

More often than not, a new comer to computer science research such as an undergraduate or graduate student would naturally ask for introductory reading on the culture and philosophy in computer science. The book “Out of Their Minds: The Lives and Discoveries of 15 Great Computer Scientists” is a nice book for them.     

Multifamily malware models

When training a machine learning model, there is likely to be a tradeoff between accuracy and the diversity of the dataset. Previous research has shown that if we train a model to detect one specific malware family, we generally obtain stronger results as compared to a case where we train a single model on multiple diverse families. However, during the detection phase, it would be more efficient to have a single model that can reliably detect multiple families, rather than having to score each sample against multiple models. In this research, we conduct experiments based on byte n-gram features to quantify the relationship between the generality of the training dataset and the accuracy of the corresponding machine learning models, all within the context of the malware detection problem. We find that neighborhood-based algorithms generalize surprisingly well, far outperforming the other machine learning techniques considered.

     

Studying lifeworlds of science learning: A longitudinal study of changing ideas,contexts, and personal orientations in science learning

Abstract

This article presents results from a long‐term study of science learning that began with 6 children during their Grade 5 study of the topic light and has now followed the original participants for 18 years. This report features details of the story of 1 participant, Donnie, and provides information on the approach to the larger program of research. The research takes three main thrusts: First, it studies how concepts in science, in this case, light, are developed over a lifetime of learning. Second, it studies the nature of the experience of science learning in school and now into the adult lives of participants. Third, the research seeks to understand the nature of changes in personal orientations to science learning over the years, from elementary education to adult life, a construct developed in the original case studies to describe features of each individual's approach to learning science (Shapiro, 1994a). Research insights highlight concept development and the importance of building self‐reflection opportunities into the curriculum and address the career and lifetime impacts of school‐science learning experience.     

A review of predictive policing from the perspective of fairness

Machine Learning has become a popular tool in a variety of applications in criminal justice, including sentencing and policing. Media has brought attention to the possibility of predictive policing systems causing disparate impacts and exacerbating social injustices. However, there is little academic research on the importance of fairness in machine learning applications in policing. Although prior research has shown that machine learning models can handle some tasks efficiently, they are susceptible to replicating systemic bias of previous human decision-makers. While there is much research on fair machine learning in general, there is a need to investigate fair machine learning techniques as they pertain to the predictive policing. Therefore, we evaluate the existing publications in the field of fairness in machine learning and predictive policing to arrive at a set of standards for fair predictive policing. We also review the evaluations of ML applications in the area of criminal justice and potential techniques to improve these technologies going forward. We urge that the growing literature on fairness in ML be brought into conversation with the legal and social science concerns being raised about predictive policing. Lastly, in any area, including predictive policing, the pros and cons of the technology need to be evaluated holistically to determine whether and how the technology should be used in policing.

     

近十年（2008-2017 年）大数据下CDIO 理念与计算机学科融合研究动态

CDIO的核心理念是“做中学”和“项目化教学”,基于CDIO的工程理念展开对计算机专业人才的培养和学科发展的研究越来越受到教育者和研究者的关注。因此,本文基于知网的可视化分析能力通过关键词索引探究近十年来在CDIO理念引导下的计算机学科专业研究趋势与发展方向。     

Being your own role model for improving self‐efficacy: An elementary teacher self‐actualizes through drama‐based science teaching

Abstract:

Many teachers in elementary schools lack school science self‐efficacy, largely because of their inexperience with the subject. This frequently leads them to avoid teaching science or to teach it in ways that compromise the development of aspects of students’ scientific literacy. This paper describes how one teacher was able to improve her school science self‐efficacy through facilitated action research. In response to becoming aware of a discrepancy between her school science practices and her fundamental educational beliefs, Lisa developed a drama‐based, integrated science unit that she judged successful in helping students to achieve relevant learning goals. This experience led Lisa and her students to feel much more positive about teaching and learning in school science. Rather than learning from another, however, “Lisa, the science teacher” learned— to a great extent—from “Lisa, the drama‐based educator.” This finding has implications for science‐phobic teachers and for facilitators of their action research.     

Expert systems for knowledge management: crossing the chasm between information processing and sense making

Based on insights from research in information systems, information science, business strategy and organization science, this paper develops the bases for advancing the paradigm of AI and expert systems technologies to account for two related issues: (a) dynamic radical discontinuous change impacting organizational performance; and (b) human sense-making processes that can complement the machine learning capabilities for designing and implementing more effective knowledge management systems.     

流形学习中的算法研究

详细介绍了一种新的机器学习的方法--流形学习.流形学习是一种新的非监督学习方法,可以有效地发现高维非线性数据集的内在维数并进行维数约简,近年来越来越受到机器学习和认知科学领域的研究者的重视.目前已经出现了很多有效的流形学习算法,如等度规映射(ISOMAP)、局部线性嵌套(Locally Linear Embedding ,LLE)等.详细讲述了当前常用的几种流形学习算法以及在流形方面已经取得的研究成果,并对流形学习目前在各方面的应用作了较为细致的阐述.最后展望了流形学习的研究发展趋势,且提出了流形学习中仍需解决的关键问题.     

Generalizing predicates with string arguments

The least general generalization (LGG) of strings may cause an over-generalization in the generalization process of the clauses of predicates with string arguments. We propose a specific generalization (SG) for strings to reduce over-generalization. SGs of strings are used in the generalization of a set of strings representing the arguments of a set of positive examples of a predicate with string arguments. In order to create a SG of two strings, first, a unique match sequence between these strings is found. A unique match sequence of two strings consists of similarities and differences to represent similar parts and differing parts between those strings. The differences in the unique match sequence are replaced to create a SG of those strings. In the generalization process, a coverage algorithm based on SGs of strings or learning heuristics based on match sequences are used. Ilyas Cicekli received a Ph.D. in computer science from Syracuse University in 1991. He is currently a professor of the Department of Computer Engineering at Bilkent University. From 2001 till 2003, he was a visiting faculty at University of Central Florida. His current research interests include example-based machine translation, machine learning, natural language processing, and inductive logic programming. Nihan Kesim Cicekli is an Associate Professor of the Department of Computer Engineering at the Middle East Technical University (METU). She graduated in computer engineering at the Middle East Technical University in 1986. She received the MS degree in computer engineering at Bilkent University in 1988; and the PhD degree in computer science at Imperial College in 1993. She was a visiting faculty at University of Central Florida from 2001 till 2003. Her current research interests include multimedia databases, semantic web, web services, data mining, and machine learning.     

Solving multi-instance problems with classifier ensemble based on constructive clustering

In multi-instance learning, the training set is composed of labeled bags each consists of many unlabeled instances, that is, an object is represented by a set of feature vectors instead of only one feature vector. Most current multi-instance learning algorithms work through adapting single-instance learning algorithms to the multi-instance representation, while this paper proposes a new solution which goes at an opposite way, that is, adapting the multi-instance representation to single-instance learning algorithms. In detail, the instances of all the bags are collected together and clustered into d groups first. Each bag is then re-represented by d binary features, where the value of the ith feature is set to one if the concerned bag has instances falling into the ith group and zero otherwise. Thus, each bag is represented by one feature vector so that single-instance classifiers can be used to distinguish different classes of bags. Through repeating the above process with different values of d, many classifiers can be generated and then they can be combined into an ensemble for prediction. Experiments show that the proposed method works well on standard as well as generalized multi-instance problems. Zhi-Hua Zhou is currently Professor in the Department of Computer Science & Technology and head of the LAMDA group at Nanjing University. His main research interests include machine learning, data mining, information retrieval, and pattern recognition. He is associate editor of Knowledge and Information Systems and on the editorial boards of Artificial Intelligence in Medicine, International Journal of Data Warehousing and Mining, Journal of Computer Science & Technology, and Journal of Software. He has also been involved in various conferences. Min-Ling Zhang received his B.Sc. and M.Sc. degrees in computer science from Nanjing University, China, in 2001 and 2004, respectively. Currently he is a Ph.D. candidate in the Department of Computer Science & Technology at Nanjing University and a member of the LAMDA group. His main research interests include machine learning and data mining, especially in multi-instance learning and multi-label learning.