William F. Punch

• PhD
• Professor (Associate) at Michigan State University

Genetic programming has been successfully applied to several real-world problem domains. One such application area is image classification, wherein genetic programming has been used for a variety of problems such as breast cancer detection, face detection, and pedestrian detection, to name a few. We present the use of genetic programming for detect...

Premature convergence is a serious problem that plagues genetic programming, stifling its search performance. Several genetic diversity maintenance techniques have been proposed for combating premature convergence and improving search efficiency in genetic programming. Recent research has shown that while genetic diversity is important, focusing di...

Many diversity techniques have been developed for addressing premature convergence, which is a serious problem that stifles the search effectiveness of evolutionary algorithms. However, approaches that aim to avoid premature convergence can often take longer to discover a solution. The Genetic Marker Diversity algorithm is a new technique that has...

Examining the properties of local optima is a common method for understanding combinatorial-problem landscapes. Unfortunately, exhaustive algorithms for finding local optima are limited to very small problem sizes. We propose a method for exploiting problem structure to skip hyperplanes that cannot contain local optima, allowing runtime to scale wi...

Unlike black-box optimization problems, gray-box optimization problems have known, limited, non-linear relationships between variables. Though more restrictive, gray-box problems include many real-world applications in network security, computational biology, VLSI design, and statistical physics. Leveraging these restrictions, the Hamming-Ball Hill...

Genetic diversity plays an important role in avoiding premature convergence, which is a phenomenon that stifles the search effectiveness of evolutionary algorithms. However, approaches that avoid premature convergence by maintaining genetic diversity can do so at the cost of efficiency, requiring more fitness evaluations to find high quality soluti...

The Parameter-less Population Pyramid (P3) is a recently introduced method for performing evolutionary optimization without requiring any user specified parameters. P3's primary innovation is to replace the generational model with a pyramid of multiple populations that are iteratively created and expanded. In combination with local search and advan...

Data centers have become a critical computing infrastructure in the era of cloud computing. Temperature monitoring and forecasting are essential for preventing overheating-induced server shutdowns and improving a data center's energy efficiency. This paper presents a novel cyber-physical approach for temperature forecasting in data centers, which i...

There is little doubt in scientific circles that--counting from the origin of life towards today--evolution has led to an increase in the amount of information stored within the genomes of the biosphere. This trend of increasing information on average likely holds for every successful line of descent, but it is not clear whether this increase is du...

Real world applications of evolutionary techniques are often hindered by the need to determine problem specific parameter settings. While some previous methods have reduced or removed the need for parameter tuning, many do so by trading efficiency for general applicability. The Parameter-less Population Pyramid (P3) is an evolutionary technique tha...

Clustering ensembles combine multiple partitions of data into a single clustering solution of better quality. Inspired by the success of supervised bagging and boosting algorithms, we propose non-adaptive and adaptive resampling schemes for the integration of multiple independent and dependent clusterings. We investigate the effectiveness of baggin...

Understanding how search operators interact with solution representation is a critical step to improving search. In Cartesian genetic programming (CGP), and genetic programming (GP) in general, the complex genotype to phenotype map makes achieving this understanding a challenge. By examining aspects such as tuned parameter values, the search qualit...

The Institute for Cyber-Enabled Research (iCER) at Michigan State University (MSU) was established in 2009 to coordinate and support multidisciplinary resources for computation and computational sciences. iCER is the home of MSU's centralized High Performance Computing resources, which include a heterogeneous compute cluster with various hardware d...

In this paper we examine how Cartesian Genetic Programming's (CGP's) method for encoding directed acyclic graphs (DAGs) and its mutation operator bias the effective length of individuals as well as the distribution of inactive nodes in the genome. We investigate these biases experimentally using two CGP variants as comparisons: Reorder, a method fo...

Cartesian Genetic Programming (CGP) is a form of Genetic Programming (GP) where a large proportion of the genome is identifiably unused by the phenotype. This can lead mutation to create offspring that are genotypically different but phenotypically identical, and therefore do not need to be evaluated. We investigate theoretically and empirically th...

Fault localization (FL) is the process of debugging erroneous code and directing analysts to the root cause of the bug. With this in mind, we have developed a distributed, end-to-end fuzzing and analysis system that starts with a binary, identifies bugs, and subsequently localizes the bug's root cause. Our system does not require the test subject's...

Faced with a fragmented research computing environment and growing needs for high performance computing resources, Michigan State University established the High Performance Computing Center in 2005 to serve as a central high performance computing resource for MSU's research community. Like greenfield industrial development, the center was unconstr...

Data centers have become a critical computing infrastructure in the era of cloud computing. Temperature monitoring and forecasting are essential for preventing overheating-induced server shutdowns and improving a data center's energy efficiency. This paper presents a novel cyber-physical approach for temperature forecasting in data centers, which i...

The ubiquity of multicore computers and various architectures that these computers employ make training in High Performance Computing (HPC) a necessity for modern programmers. In fact, there is good reason to think that introductory programmers should be introduced to the concepts of multicore very early in their education. The question is how to d...

This paper describes a generic, knowledge-based mechanism for selecting among a fixed set of alternatives. The mechanism, termed sponsor-selector has been used as a control mechanism in a number of different knowledge-based systems including problem-solver integration applications in routine design, diagnostic problem-solving, and navigational plan...

Incorporating semantic knowledge from an ontology into document clustering is an important but challenging problem. While numerous methods have been developed, the value of using such an ontology is still not clear. We show in this paper that an ontology can be used to greatly reduce the number of features needed to do document clustering. Our hypo...

Several approaches have been introduced for modeling and prediction of nonlinear dynamics which have chaotic characteristics. Among these methods, data driven approaches such as Auto Regressive (AR) models, Nonlinear Auto Regressive (NAR) models, Radial Basis Function (RBF) networks, and Multi Layered Perceptron (MLP) neural networks have proven th...

A continuous and fully automated software exploit discovery and development pipeline for real-world problems has not yet been achieved, but is desired by defenders and attackers alike. We have made significant steps toward that goal by combining and enhancing known bug hunting and analysis techniques. The first step is the implementation of an easy...

If you change the CS1 language to Python, what is the impact on the rest of the curriculum? In earlier work we examined the impact of changing CS1 from C++ to Python while leaving CS2 in C++. We found that Python-prepared CS1 students fared no differently in CS2 than students whose CS1 course was in C++, even though CS2 was taught in C++ and covere...

We recently converted a CS1 (Introduction to Computing) class to use the Python language in place of C++. Among other reasons, we hoped that the new language would help students who typically struggled with the course. Our typical drop+fail rate was around 25%-30% for C++, and we hoped the conversion would reduce this number. Though it did reduce s...

How suitable is a Python-based CS1 course as preparation for a C++-based CS2 course? After fifteen years of using C++ for both CS1 and CS2, the Computer Science Depart- ment at Michigan State University changed the CS1 course to Python. This paper examines the impact of that change on the second course in the sequence, CS2, which kept C++ as its pr...

Incorporating background knowledge into data mining algorithms is an important but challenging problem. Current approaches in semi-supervised learning require explicit knowledge provided by domain experts, knowledge specific to the particular data set. In this study, we propose an ensemble model that couples two sources of information: statistics i...

Computational techniques for nanostructure determination of substances that resist standard crystallographic methods are often laborious processes starting from initial guess solutions not derived from experimental data. The Liga algorithm can create nanostructures using only lists of lengths or distances between atom pairs, providing an experiment...

This paper presents an algorithm for recommending items using a diverse set of features. The items are recommended by performing a random walk on the k-partite graph constructed from the heterogenous features. To support personalized recommendation, the random walk must be initiated separately for each user, which is computationally demanding given...

This paper is aimed to describe a general improvement over the previous work on the cooperative multiagent coordination. The focus is on highly dynamic environments where the message transfer delay is not negligible. Therefore, the agents shall not count on communicating their intentions along the time they are making the decisions, because this w...

Advances in materials science and molecular biology followed rapidly from the ability to characterize atomic structure using single crystals. Structure determination is more difficult if single crystals are not available. Many complex inorganic materials that are of interest in nanotechnology have no periodic long-range order and so their structure...

Spatial based gene selection for division of chromosomes used by crossover operators is proposed for three-dimensional problems. This spatial selection is shown to preserve more genetic material and reduce the disruptive effects of crossover. The disruptive effects of crossover can be quantified by count- ing the destruction of subgraphs that repre...

Clustering ensembles have emerged as a powerful method for improving both the robustness as well as the stability of unsupervised classification solutions. However, finding a consensus clustering from multiple partitions is a difficult problem that can be approached from graph-based, combinatorial, or statistical perspectives. This study extends pr...

A meta-GA (GA within a GA) is used to investigate evolving the parameter settings of genetic operators for genetic and evolutionary algorithms (GEA) in the hope of creating a self- adaptive GEA. We report three findings. First, the meta-GA can adapt its genetic operators to different problems and thereby perform well on average across diverse probl...

A genetic algorithm is proposed with real value variables, spatially based crossover operator, a small mutation, large scale mutation, vector sum local search and geometric only based objective function to generate candidate molecule conformations from atomic pair distance data. To better simulate experimental data only information from the pair di...

An important goal of data mining is to discover the unobvious relationships among the objects in a data set. Web-based educational systems collect vast amounts of data on user patterns, and data mining methods can be applied to these databases to discover interesting associations between student attributes, problem attributes, and solution strategi...

We present two programs: gafs for optimal selection of loci for use in individual assignment tests, and mlc, a program for individual classification using maximum likelihood and k-nearest neighbour decision rules. gafs software employs a genetic algorithm to heuristically search multilocus subsets with several objective functions to maximize predic...

Classification fusion combines multiple classifications of data into a single classification solution of greater accuracy. Feature extraction aims to re- duce the computational cost of feature measurement, increase classifier effi- ciency, and allow greater classification accuracy based on the process of deriv- ing new features from the original fe...

Clustering ensembles have emerged as a powerful method for improving both the robustness and the stability of unsupervised classification solutions. However, finding a consensus clustering from multiple partitions is a difficult problem that can be approached from graph-based, combinatorial or statistical perspectives. We offer a probabilistic mode...

Synopsis We demonstrate the effectiveness of a genetic algorithm for discovering multi-locus combinations that provide accurate individual assignment decisions and estimates of mixture composition based on likelihood classification. Using simulated data representing different levels of inter-population differentiation (F st ∼ 0.01 and 0.10), geneti...

Combination of multiple clusterings is an important task in the area of unsupervised learning. Inspired by the success of supervised bagging algorithms, we propose a resampling scheme for integration of multiple independent clusterings. Individual partitions in the ensemble are sequentially generated by clustering specially selected subsamples of t...

Clustering ensembles combine multiple partitions of the given data into a single clustering solution of better quality. Inspired by the success of supervised boosting algorithms, we devise an adaptive scheme for integration of multiple non-independent clusterings. Individual partitions in the ensemble are sequentially generated by clustering specia...

Recently web-based educational systems collect vast amounts of data on user patterns, and data mining methods can be applied to these databases to discover interesting associations based on students' features and the actions taken by students in solving homework and exam problems. The main purpose of data mining is to discover the hidden relationsh...

Web-based educational technologies allow educators to study how students learn (descriptive studies) and which learning strategies are most effective (causal/predictive studies). Since web-based educational systems collect vast amounts of student profile data, data mining and knowledge discovery techniques can be applied to find interesting relatio...

The combination of multiple clusterings is a difficult problem in the practice of distributed data mining. Both the cluster generation mechanism and the partition integration process influence the quality of the combinations. We propose a data resampling approach for building cluster ensembles that are both robust and stable. In particular, we inve...

A data set can be clustered in many ways depending on the clustering algorithm employed, parameter settings used and other factors. Can multiple clusterings be combined so that the final partitioning of data provides better clustering? The answer depends on the quality of clusterings to be combined as well as the properties of the fusion method. Fi...

Newly developed Web-based educational technologies offer researchers unique opportunities to study how students learn and what approaches to learning lead to success. Web-based systems routinely collect vast quantities of data on user patterns, and data mining methods can be applied to these databases. This paper presents an approach to classifying...

Feature extraction based on evolutionary search offers new possibili- ties for improving,classification accuracy and reducing measurement,complex- ity in many data mining and machine learning applications. We present a family ofgenetic algorithms for feature synthesis through clustering of discrete attrib- ute values. The approach ,uses new ,compac...

This paper presents an approach for classifying students in order to predict their final grade based on features extracted from logged data in an edu- cation web-based system. A combination of multiple classifiers leads to a sig- nificant improvement in classification performance. Through weighting the fea- ture vectors using a Genetic Algorithm we...

this article is to describe several of the more widely used machine learning classifiers that may have utility when used with empirical population genetics data. We compare likelihoodbased "assignment tests" (Paetkau et al. 1995) with supervised machine learning classifiers including ANN, decision tree, and a k-NN clustering. Simulations were condu...

A key element of many bioinformatics research problems is the extraction of meaningful information from large experimental data sets. Various approaches, including statistical and graph theoretical methods, data mining, and computational pattern recognition, have been applied to this task with varying degrees of success. We have previously shown th...

A key element of bioinformatics research is the extraction of meaningful information from large experimental data sets. Various approaches, including statistical and graph theoretical methods, data mining, and computational pattern recognition, have been applied to this task with varying degrees of success. Using a novel classifier based on the Bay...

Information technology in education has greatly enhanced feedback for students and instructors. However, the feedback for instructors has been more difficult to get as individualization of problems has evolved in sophistication and complexity to insure that students are inhibited from mindless copying other students' work without impeding useful co...

Classification methods used in machine learning (e.g., artificial neural networks, decision trees, and k-nearest neighbor clustering) are rarely used with population genetic data. We compare different nonparametric machine learning techniques with parametric likelihood estimations commonly employed in population genetics for purposes of assigning i...

Information technology in education has greatly enhanced feedback for students and instructors. However, the feedback for instructors has been more difficult to get as individualization of problems has evolved in sophistication and complexity to insure that students are inhibited from mindless copying other students' work without impeding useful co...

A key element of bioinformatics research is the extraction of meaningful information from large experimental data sets. Various approaches, including statistical and graph theoretical methods, data mining, and computational pattern recognition, have been applied to this task with varying degrees of success. We have previously shown that a genetic a...

Introduction New optimization problems arise every day -- for instance, what is the quickest path to work? Where and how congested is the road construction? Am I better off riding my bike? If so, what is the shortest path? Sometimes these problems are easily solved, but many engineering problems cannot be handled satisfactorily using traditional op...

A key element of many bioinformatics research problems is the extraction of meaningful information from large experimental data sets. Various approaches, includ- ing statistical and graph theoretical methods, data mining, and computational pattern recognition, have been applied to this task with varying degrees of success. We have pre- viously show...

We examine the effectiveness of gradient search optimization of numeric leaf values for Genetic Programming. Genetic search for tree-like programs at the population level is complemented by the optimization of terminal values at the individual level. Local adaptation of individuals is made easier by algorithmic differentiation. We show how conventi...

Pattern recognition generally requires that objects be described in terms of a set of measurable features. The selection and quality of the features representing each pattern affect the success of subsequent classification. Feature extraction is the process of deriving new features from original features to reduce the cost of feature measurement, i...

The parallel execution of several populations in evolutionary algorithms has usually given good results. Nevertheless, researchers have to date drawn conflicting conclusions when using some of the parallel genetic programming models. One aspect of the conflict is population size, since published GP works do not agree about whether to use large or s...

The parallel execution of several populations in Evolutionary Algorithms has usually given good results. Nevertheless, researchers have to date drawn conflicting conclusions when using some of the Parallel Genetic Programming models. One aspect of the conflict is population size, since published GP works do not agree about whether to use large or s...

This paper presents an approach to optimal design of elastic flywheels using an Injection Island Genetic Algorithm (iiGA), summarizing a sequence of results reported in earlier publications. An iiGA in combination with a structural finite element code is used to search for shape variations and material placement to optimize the Specific Energy Dens...

This paper generalizes our research on parameter interdependencies in reinforcement learning algorithms for optimization problem solving. This generalization expands the work to a larger class of search algorithms that use explicit search statistics to form feasible solutions. Our results suggest that genetic algorithms can both enrich and benefit...

This paper presents two approaches that address the problems of the local character of the search and imprecise state representation of reinforcement learning (RL) algorithms for solving combinatorial optimization problems. The first, Bayesian, approach aims to capture solution parameter interdependencies. The second approach combines local informa...

This paper presents an approach that uses reinforcement learning (RL) algorithms to solve combinatorial optimization problems. In particular, the approach combines both local and global search characteristics: local information as encoded by typical RL schemes and global information as contained in a population of search agents. The effectiveness o...

Evolutionary programming (EP) has historically used a number of approaches for selection of the mutation step size. Current EP implementations typically use self-adaptive meta-parameters for mutation step size selection. However, one of the potential drawbacks of this scheme is that it is not directly responsive to the variance reduction caused by...

This paper presents two approaches that address the problems of the local character of the search and imprecise state representation of reinforcement learning (RL) algorithms for solving combinatorial optimization problems. The first, Bayesian, approach aims to capture solution parameter interdependencies. The second approach combines local informa...

We describe a system for extracting concepts from unstructured text. We do this by clustering document words and then assembling a structure which relates these words semantically. The clustering process identifies words which co--occur across a set of documents and creates groups of words which suggest a semantic context common across the document...

. This paper summarizes our research on feature selection and extraction from high-dimensionality data sets using genetic algorithms. We have developed a GA-based approach utilizing a feedback linkage between feature evaluation and classification. That is, we carry out feature selection or feature extraction simultaneously with classifier design, t...

We describe a system for extracting concepts from unstructured text. We do this by identifying relationships between words in the text based on a lexical database and identifying groups of these words which form closely tied conceptual groups. The word relationships are used to create a directed graph, called a Semantic Relationship Graph (SRG). Th...

Traditional approaches to real-valued function opt imization using evolutionary computational methods tend to use eith er self-adaptive operators (as in the case of evolutionary programming), or po pulation-based operators (as in the case of most real-valued genetic algorithms) . However, in general, most population-based operators are limited in s...

Many black box optimization algorithms have sufficient flexibility to allow them to adapt to the varying circumstances they encounter. These capabilities are of two primary sorts: 1) user-determined choices among alternative parameters, operations, and logic structures, and 2) the algorithm-determined alternative paths chosen during the process of...

This paper first describes optimal design of elastic flywheels using an Injection Island Genetic Algorithm (iiGA). An iiGA in combination with a finite element code is used to search for shape variations to optimize the Specific Energy Density of flywheels (SED is the rotational energy stored per unit mass). iiGA's seek solutions simultaneously at...

The use of multiple populations in Genetic Programming is an area that is just beginning to be investigated. To date a number of conflicting reports have been generated with respect to the effectiveness of multiple populations in GP. We report here that these conflicting reports may be due a problem-dependent nature found in GP that has not been re...

Many black box optimization algorithms have sufficient flexibility to allow them to adapt to the varying circumstances they encounter. These capabilities are of two primary sorts: user-determined choices among alternative parameters, operations, and logic structures; and the algorithm-determined alternative paths chosen during the process of seekin...

This paper presents an approach to optimal design of elastic flywheels using an Injection Island Genetic Algorithm (iiGA). An iiGA in combination with a finite element code is used to search for shape variations to optimize the Specific Energy Density (SED) of elastic flywheels. SED is defined as the amount of rotational energy stored per unit mass...

We examine techniques that “discover” features in sets of pre-categorized documents, such that similar documents can be found on the World Wide Web. First, we examine techniques which will classify training examples with high accuracy, then explain why this is not necessarily useful. We then describe a method for extracting word clusters from the r...

This paper describes a GA for job shop scheduling problems. Using the Giffler and Thompson algorithm, we created two new operators, THX crossover and mutation, which better transmit temporal relationships in the schedule. The approach produced excellent results on standard benchmark job shop scheduling problems. We further tested many models and sc...

Water-mediated ligand interactions are essential to biological processes, from product displacement in thymidylate synthase to DNA recognition by Trp repressor, yet the structural chemistry influencing whether bound water is displaced or participates in ligand binding is not well characterized.Consolv, employing a hybrid k-nearest-neighbors classif...

Statistical pattern recognition techniques classify objects in terms of a representative set of features. The selection of features to measure and include can have a significant effect on the cost and accuracy of an automated classifier. Our previous research has shown that a hybrid between a k-nearest-neighbors (knn) classifier and a genetic algor...

A GA's performance on a specific problem is related to many factors, such as genetic operators and corresponding parameter settings and the representation of the problem on the chromosome. Optimization of these factors to improve the speed and robustness of search is essential to successful application of a GA. The work reported here uses a two-lev...

Blackbox Optimization(BBO) algorithms are candidate methods when knowledge of the problem is too incomplete to allow development of an efficient heuristic algorithm. Many BBOs have sufficient flexibility to allow them to adapt to the varying circumstances they encounter. These flexibilities include user-determined choices among alternative paramete...

Genetic Algorithms (GAs) are a powerful technique for search and optimization problems, and are particularly useful in the optimization of composite structures. The search space for an optimal composite structure is gener-ally discontinuous and strongly multimodal, with the possibility for many local sub-optimal solutions or even sin-gular extrema....

Finding the 3-D geometry or tertiary structure of an arbitrary protein is vital to understanding the functionality of that protein. The prediction of this structure, known as the protein folding problem, is very difficult and has been labeled one of the "grand challenge problems" for the scientific community. We report here on further work to deter...

IPCA (Intelligent Process Control Architecture) is an architecture for flexible, intelligent control. IPCA is rooted in the generic task (GT) approach to knowledge-based systems and incorporates plan generation and real-time plan execution monitoring components. The plan generation component produces a state-based process control plan representatio...

William F. Punch. A Diagnosis System Using a Task Integrated Problem Solver Architecture (TIPS), Including Causal Reasoning. PhD thesis, The Ohio State University, 1989. [18] R. Reiter. A theory of diagnosis from first principles. Artificial Intelligence, 32(1):57 95, 1987. [19] Andee Rubin. The role of hypotheses in medical diagnosis. In IJCAI 197...