Portfolio configurators tune an algorithm's parameters to produce a portfolio of parameterisations, with complementary strengths on different problem instances. However, as the portfolio size increases, the marginal contribution of each additional configuration decreases, making portfolio configuration computationally expensive. We propose a hybrid racing method that combines statistical testing (to early eliminate poorly performing configurations) and successive halving (which progressively allocates more resources to promising configurations by discarding fractions of the configuration candidate set in rounds). We connect greedy portfolio configuration approaches to the theory of submodular function optimisation, which implies an approximation ratio of 1 − 1/e for the problem in idealised form. We experimentally compare the hybrid method with pure statistical racing and pure successive halving, on algorithm configuration benchmarks from the literature. We show that the proposed hybrid racing method achieves portfolios with performance matching those configured on the full evaluation data, while usually using roughly 5% of the runs of a full evaluation, and that the method is more efficient than pure statistical racing or successive halving for portfolio configuration. Our experiments demonstrate that adaptive elimination strategies can significantly improve the efficiency of portfolio configuration methods.

The quantum approximate optimization algorithm (QAOA) tackles combinatorial optimization problems in a quantum computing context, where achieving globally optimal and exact solutions is not always feasible because of classical computational constraints or problem complexity. The performance of QAOA generally depends on finding sufficiently good parameters that facilitate competitive approximate solutions. However, this is fraught with challenges, such as “barren plateaus,” making the search for effective parameters a nontrivial endeavor. More recently, the question of whether such an optimal parameter search is even necessary has been posed, with some studies showing that optimal parameters tend to be concentrated on certain values for specific types of problem instances. However, these existing studies have only examined specific instance classes of Maximum Cut, so it is uncertain if the claims of instance independence apply to a diverse range of instances. In this paper, we use instance space analysis to study QAOA parameter initialization strategies for the first time, providing a comprehensive study of how instance characteristics affect the performance of initialization strategies across a diverse set of graph types and weight distributions. Unlike previous studies that focused on specific graph classes (e.g., d-regular or Erdős–Rényi), our work examines a much broader range of instance types, revealing insights about parameter transfer between different graph classes. We introduce and evaluate a new initialization strategy, quantum instance-based parameter initialization, that leverages instance-specific information, demonstrating its effectiveness across various instance types. Our analysis at higher QAOA depths (p = 15) provides insights into the effectiveness of different initialization strategies beyond the low-depth circuits typically studied. History: Accepted by Giacomo Nannicini, Area Editor for Quantum Computing and Operations Research. This article is accepted for Special Issue. Funding: This research was supported by the Australian Research Council [Grant IC200100009 for the Australian Research Council Training Centre in Optimization Technologies, Integrated Methodologies and Applications]. V. Katial is supported by The University of Melbourne [Research Training Program Scholarship]. The authors gratefully acknowledge the information technology infrastructure support provided by The University of Melbourne’s Research Computing Services and the Petascale Campus Initiative. This research was also supported by The University of Melbourne through the establishment of the IBM Quantum Network Hub at the university. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2024.0564 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2024.0564 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .

State-of-the-art lane detection methods often rely on specific knowledge about lanes – such as straight lines and parametric curves – to detect lane lines. While the specific knowledge can ease the modeling process, it poses challenges in handling lane lines with complex topologies (e.g., dense, forked, curved, etc.). Recently, dynamic convolution-based methods have shown promising performance by utilizing the features from some key locations of a lane line, such as the starting point, as convolutional kernels, and convoluting them with the whole feature map to detect lane lines. While such methods reduce the reliance on specific knowledge, the kernels computed from the key locations fail to capture the lane line’s global structure due to its long and thin structure, leading to inaccurate detection of lane lines with complex topologies. In addition, the kernels resulting from the key locations are sensitive to occlusion and lane intersections. To overcome these limitations, we propose a transformer-based dynamic kernel generation architecture for lane detection. It utilizes a transformer to generate dynamic convolutional kernels for each lane line in the input image, and then detect these lane lines with dynamic convolution. Compared to the kernels generated from the key locations of a lane line, the kernels generated with the transformer can capture the lane line’s global structure from the whole feature map, enabling them to effectively handle occlusions and lane lines with complex topologies. We evaluate our method on three lane detection benchmarks, and the results demonstrate its state-of-the-art performance. Specifically, our method achieves an F1 score of 63.40 on OpenLane and 88.47 on CurveLanes, surpassing the state of the art by 4.30 and 2.37 points, respectively.

Item Response Theory (IRT) has been proposed within the field of Educational Psychometrics to assess student ability as well as test question difficulty and discrimination power. More recently, IRT has been applied to evaluate machine learning algorithm performance on a single classification dataset, where the student is now an algorithm, and the test question is an observation to be classified by the algorithm. In this paper we present a modified IRT-based framework for evaluating a portfolio of algorithms across a repository of datasets, while simultaneously eliciting a richer suite of characteristics - such as algorithm consistency and anomalousness - that describe important aspects of algorithm performance. These characteristics arise from a novel inversion and reinterpretation of the traditional IRT model without requiring additional dataset feature computations. We test this framework on algorithm portfolios for a wide range of applications, demonstrating the broad applicability of this method as an insightful algorithm evaluation tool. Furthermore, the explainable nature of IRT parameters yield an increased understanding of algorithm portfolios.

Search-based software testing (SBST) is now a mature area, with numerous techniques developed to tackle the challenging task of software testing. SBST techniques have shown promising results and have been successfully applied in the industry to automatically generate test cases for large and complex software systems. Their effectiveness, however, has been shown to be problem dependent. In this paper, we revisit the problem of objective performance evaluation of SBST techniques in light of recent methodological advances – in the form of Instance Space Analysis (ISA) – enabling the strengths and weaknesses of SBST techniques to be visualised and assessed across the broadest possible space of problem instances (software classes) from common benchmark datasets. We identify features of SBST problems that explain why a particular instance is hard for an SBST technique, reveal areas of hard and easy problems in the instance space of existing benchmark datasets, and identify the strengths and weaknesses of state-of-the-art SBST techniques. In addition, we examine the diversity and quality of common benchmark datasets used in experimental evaluations.

In recent years, multifidelity expensive black-box (Mf-EBB) methods have received increasing attention due to their strong applicability to industrial design problems. The challenge, however, is that knowledge of the relationship between decisions and objective values is limited to a small set of sample observations of variable quality. In the field of Mf-EBB, a problem instance consists of an expensive yet accurate source of information, and one or more cheap yet less accurate sources of information. The field aims to provide techniques either to accurately explain how decisions affect design outcome, or to find the best decisions to optimise design outcomes. Many techniques that use surrogate models have been developed to provide solutions to both aims. Only in recent years, however, have researchers begun to explore the conditions under which these new techniques are reliable, often focusing on problems with a single low-fidelity function, known as bifidelity expensive black-box (Bf-EBB) problems. This study extends the existing Bf-EBB test instances found in the literature, as well as the features used to determine when the low-fidelity information source should be used. A literature test suite is constructed and augmented with new instances to demonstrate the potentially misleading results that could be reached using only the instances currently found in the literature, and to expose the criticality of a more heterogeneous test suite for algorithm assessment. Addressing the shortcomings of the existing literature, a new set of features is presented, as well as a new instance creation procedure, and a study of their impact on algorithm assessment is conducted. The low-fidelity information source is shown to be valuable if it is often locally accurate, even when its overall accuracy is relatively low. This contradicts the existing literature guidelines, which indicate the low-fidelity information is only useful if it has a high overall accuracy. History: Accepted by Antonio Frangioni, Area Editor for Design & Analysis of Algorithms – Continuous. Funding: This work was supported by Australian Research Council [Grant IC200100009] for the ARC Training Centre in Optimisation Technologies, Integrated Methodologies and Applications (OPTIMA), and the University of Melbourne Research Computing Services and Petascale Campus Initiative. N. Andrés-Thió is also supported by a Research Training Program scholarship from the University of Melbourne. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplementary Information [ https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2022.1217 ] or is available from the IJOC GitHub software repository ( https://github.com/INFORMSJoC ) at [ http://dx.doi.org/10.5281/zenodo.6578060 ].

Machine Learning studies often involve a series of computational experiments in which the predictive performance of multiple models are compared across one or more datasets. The results obtained are usually summarized through average statistics, either in numeric tables or simple plots. Such approaches fail to reveal interesting subtleties about algorithmic performance, including which observations an algorithm may find easy or hard to classify, and also which observations within a dataset may present unique challenges. Recently, a methodology known as Instance Space Analysis was proposed for visualizing algorithm performance across different datasets. This methodology relates predictive performance to estimated instance hardness measures extracted from the datasets. However, the analysis considered an instance as being an entire classification dataset and the algorithm performance was reported for each dataset as an average error across all observations in the dataset. In this paper, we developed a more fine-grained analysis by adapting the ISA methodology. The adapted version of ISA allows the analysis of an individual classification dataset by a 2-D hardness embedding, which provides a visualization of the data according to the difficulty level of its individual observations. This allows deeper analyses of the relationships between instance hardness and predictive performance of classifiers. We also provide an open-access Python package named PyHard, which encapsulates the adapted ISA and provides an interactive visualization interface. We illustrate through case studies how our tool can provide insights about data quality and algorithm performance in the presence of challenges such as noisy and biased data.

When demonstrating the effectiveness of a new algorithm, researchers are traditionally encouraged to compare their algorithm's performance against existing algorithms on well-studied benchmark test suites. In the absence of more nuanced methodologies, algorithm performance is typically summarized on average across the test suite examples. This paper highlights the potential bias of conclusions drawn by analyzing "on average" performance, and the opportunities offered by a recent testing methodology known as instance space analysis. To illustrate, we revisit our 2007 comparative study of algorithms for facial age estimation, and rigorously stress-test to challenge the original conclusions. The case study demonstrates how powerful visualizations offered by instance space analysis enable greater insights into unique strengths and weaknesses, and which algorithm should be used when and why. Inspired by such insights, a new algorithm is proposed, and its unique advantage is demonstrated. The bias often hidden in well-studied datasets, and the ramifications for drawing biased conclusions, are also illustrated in this case study. While focused on facial age estimation, the methodology and lessons learned from the case study are broadly applicable to any study seeking to draw conclusions about algorithm performance based on empirical results.

The role of carefully constructed benchmark suites in algorithm design and testing is critical. Within the continuous multiobjective optimization domain, existing suites include the general purpose ZDT, DTLZ, and WFG suites, and more recent ones specifically designed to explore the impacts of a particular problem characteristic. However, the relationship between existing suites is not clear, and the field would benefit from a “stock-take” assessment. This article investigates the coverage of current continuous multiobjective suites using the instance space analysis (ISA) methodology. Exploratory landscape analysis is used to measure critical features of each problem suite. Thereafter, we generate a 2-D visualization of the existing problem instances by locating them in the instance space, assessing their diversity, and identifying whether there are sparse areas of value to fill with new problem instances. Our findings show that the current suites are restricted in diversity when representing the entire problem instance space. We propose and evaluate three problem construction methods: 1) problem tuning; 2) toolkit hybridization; and 3) new function injection. Problem tuning is shown to generate problems surrounding existing instances, while hybridization creates problems falling between existing suites. Furthermore, utilizing the insights afforded by ISA, we show how problem features can be identified to inform the creation of new functions which fill gaps toward the boundaries of the instance space.

The quest for greater insights into algorithm strengths and weaknesses, as revealed when studying algorithm performance on large collections of test problems, is supported by interactive visual analytics tools. A recent advance is Instance Space Analysis, which presents a visualization of the space occupied by the test datasets, and the performance of algorithms across the instance space. The strengths and weaknesses of algorithms can be visually assessed, and the adequacy of the test datasets can be scrutinized through visual analytics. This article presents the first Instance Space Analysis of regression problems in Machine Learning, considering the performance of 14 popular algorithms on 4,855 test datasets from a variety of sources. The two-dimensional instance space is defined by measurable characteristics of regression problems, selected from over 26 candidate features. It enables the similarities and differences between test instances to be visualized, along with the predictive performance of regression algorithms across the entire instance space. The purpose of creating this framework for visual analysis of an instance space is twofold: one may assess the capability and suitability of various regression techniques; meanwhile the bias, diversity, and level of difficulty of the regression problems popularly used by the community can be visually revealed. This article shows the applicability of the created regression instance space to provide insights into the strengths and weaknesses of regression algorithms, and the opportunities to diversify the benchmark test instances to support greater insights.

In recent years, there has been a continuous stream of development in evolutionary multi-objective optimization (EMO) algorithms. The large quantity of existing algorithms introduces difficulty in selecting suitable algorithms for a given problem instance. In this paper, we perform instance space analysis on discrete multi-objective optimization problems (MOPs) for the first time under three different conditions. We create visualizations of the relationship between problem instances and algorithm performance for instance features previously identified using decision trees, as well an independent feature selection. The suitability of these features in discriminating between algorithm performance and understanding strengths and weaknesses is investigated. Furthermore, we explore the impact of various definitions of “good” performance. The visualization of the instance space provides an alternative method of algorithm discrimination by showing clusters of instances where algorithms perform well across the instance space. We validate the suitability of existing features and identify opportunities for future development.

This article presents a method to generate diverse and challenging new test instances for continuous black-box optimization. Each instance is represented as a feature vector of exploratory landscape analysis measures. By projecting the features into a two-dimensional instance space, the location of existing test instances can be visualized, and their similarities and differences revealed. New instances are generated through genetic programming which evolves functions with controllable characteristics. Convergence to selected target points in the instance space is used to drive the evolutionary process, such that the new instances span the entire space more comprehensively. We demonstrate the method by generating two-dimensional functions to visualize its success, and ten-dimensional functions to test its scalability. We show that the method can recreate existing test functions when target points are co-located with existing functions, and can generate new functions with entirely different characteristics when target points are located in empty regions of the instance space. Moreover, we test the effectiveness of three state-of-the-art algorithms on the new set of instances. The results demonstrate that the new set is not only more diverse than a well-known benchmark set, but also more challenging for the tested algorithms. Hence, the method opens up a new avenue for developing test instances with controllable characteristics, necessary to expose the strengths and weaknesses of algorithms, and drive algorithm development.

After the Big Bang, it took about 200 million years before the very first stars would form - now more than 13 billion years ago. Unfortunately, we will not be able to observe these stars directly. Instead, we can observe the 'fossil' records that these stars have left behind, preserved in the oldest stars of our own galaxy. When the first stars exploded as supernovae, their ashes were dispersed and the next generation of stars formed, incorporating some of the debris. We can now measure the chemical abundances in those old stars, which is similar to a genetic fingerprint that allows us to identify the parents.

This paper proposes a framework that provides a strategy for police to allocate resources to tackle crime, by integrating data mining models for dynamic crime prediction with a game theoretical approach to recognize the adversarial nature of the problem. The proposed framework is applied to a real case study from Santiago (Chile), and compared to other strategies involving game theory or data mining alone. The hybrid approach is demonstrated to lead to improved payoffs for the police and reduced payoffs for the criminals. A robustness analysis explores how accuracy of the data mining models affects the outcomes of the game, showing that the proposed approach can absorb significant forecasting errors while still producing superior outcomes for the police.

Automated test suite generation (ATSG) is an important topic in software engineering, with a wide range of techniques and tools being used in academia and industry. While their usefulness is widely recognized, due to the labor-intensive nature of the task, the effectiveness of the different techniques in automatically generating test cases for different software systems is not thoroughly understood. Despite many studies introducing various ATSG techniques, much remains to be learned, however, about what makes a particular technique work well (or not) for a specific software system. In this paper, we seek an answer to the question: “What features of a software system impact the effectiveness of ATSG techniques?” Once these features are identified, can they be used to select the most effective ATSG technique for a particular software system? To this end, we have implemented the mapping the effectiveness of test automation (META) tool, a new framework that identifies important software features that can be used to select suitable ATSG techniques to apply to new software systems. We evaluate the framework on a large set of open-source software projects and three ATSG techniques. The evaluation indicates that the number of methods in a class, the coupling between object classes, and the response for a class are the most indicative of what makes a software system hard to test by different techniques. The decision tree for ATSG technique selection generated by the META framework has an 88% accuracy, as shown by n-fold cross validation.

This article presents a method for the objective assessment of an algorithm's strengths and weaknesses. Instead of examining the performance of only one or more algorithms on a benchmark set, or generating custom problems that maximize the performance difference between two algorithms, our method quantifies both the nature of the test instances and the algorithm performance. Our aim is to gather information about possible phase transitions in performance, that is, the points in which a small change in problem structure produces algorithm failure. The method is based on the accurate estimation and characterization of the algorithm footprints, that is, the regions of instance space in which good or exceptional performance is expected from an algorithm. A footprint can be estimated for each algorithm and for the overall portfolio. Therefore, we select a set of features to generate a common instance space, which we validate by constructing a sufficiently accurate prediction model. We characterize the footprints by their area and density. Our method identifies complementary performance between algorithms, quantifies the common features of hard problems, and locates regions where a phase transition may lie.

The multifunction robotic cell scheduling problem has been recently studied in the literature. The main assumption in the pertaining literature is that a multifunction robot performs an operation on the part during any loaded move between two adjacent processing stages. For a two-machine cell, these stages are the input hopper, the first machine, the second machine and the output hopper. Consequently, the multifunction robot performs three operations with fixed processing times. In contrast, we assume a class of two-machine cells where none of the processing times of three operations are fixed. However, their summation is fixed and equivalent to the processing time of the unique operation. The processing mode of the unique operation performed by the multifunction robot is “stop resume.” Thus, regardless of the gap interrupts during operations by two machines, the robot continues performing the unique operation of the part when it is reloaded to the robot without any loss of time. The focus lies on n-unit cycles due to their popularity. It is proven one-unit cycles have better performance for the problem under study. The cycle time of one-unit cycles are obtained and optimality conditions are determined for different pickup criteria: free, interval, and no-wait.

Exploratory Landscape Analysis (ELA) measures have been shown to predict algorithm performance; hence, they are being applied on critical tasks such as automatic algorithm selection and problem generation. This paper provides a cautionary examination on their use in black-box continuous optimization. We explore the effect that translations have on the measures, when the cost function is defined within a bound-constrained region. Furthermore, we examine the robustness of the neighborhood structure after dimensionality reduction. The results demonstrate that a measure may transition abruptly due a translation. Therefore, we should not generalize the measures of an instance nor report average values of a measure as belonging to the generating function. Moreover, dimensionality reduction could alter the neighborhood structure, such that the regions corresponding to significantly different functions overlap.

Optimization of robotic workcells is a growing concern in automated manufacturing systems. This study develops a methodology to maximize the production rate of a multifunction robot (MFR) operating within a rotationally arranged robotic cell. An MFR is able to perform additional special operations while in transit between transferring parts from adjacent processing stages. Considering the free-pickup scenario, the cycle time formulas are initially developed for small-scale cells where an MFR interacts with either two or three machines. A methodology for finding the optimality regions of all possible permutations is presented. The results are then extended to the no-wait pickup scenario in which all parts must be processed from the input hopper to the output hopper, without any interruption either on or between machines. This analysis enables insightful evaluation of the productivity improvements of MFRs in real-life robotized workcells.

BACKGROUND: Mathematical modeling is an important tool in systems biology to study the dynamic property of complex biological systems. However, one of the major challenges in systems biology is how to infer unknown parameters in mathematical models based on the experimental data sets, in particular, when the data are sparse and the regulatory network is stochastic. RESULTS: To address this issue, this work proposed a new algorithm to estimate parameters in stochastic models using simulated likelihood density in the framework of approximate Bayesian computation. Two stochastic models were used to demonstrate the efficiency and effectiveness of the proposed method. In addition, we designed another algorithm based on a novel objective function to measure the accuracy of stochastic simulations. CONCLUSIONS: Simulation results suggest that the usage of simulated likelihood density improves the accuracy of estimates substantially. When the error is measured at each observation time point individually, the estimated parameters have better accuracy than those obtained by a published method in which the error is measured using simulations over the entire observation time period.

To study the dynamic properties of complex biological systems, mathematical modeling has been used widely in systems biology. Apart from the well-established knowledge for modeling techniques, there are still some difficulties while understanding the dynamics in system biology. One of the major challenges is how to infer unknown parameters in mathematical models based on the experimentally observed data sets. This is extremely difficult when the experimental data are sparse and the biological systems are stochastic. To tackle this problem, in this work we revised one computation method for inference called approximate Bayesian computation (ABC) and conducted extensive computing tests to examine the influence of a number of factors on the performance of ABC. Based on simulation results, we found that the number of stochastic simulations and step size of the observation data have substantial influence on the estimation accuracy. We applied the ABC method to two stochastic systems to test the efficiency and effectiveness of the ABC and obtained promising approximation for the unknown parameters in the systems. This work raised a number of important issues for designing effective inference methods for estimating rate constants in biochemical reaction systems.

A method for extracting and classifying shapes from noisy time series is proposed. The method consists of two steps. The first step is to perform a noise test on each subsequence extracted from the series using a sliding window. All the subsequences recognised as noise are removed from further analysis, and the shapes are extracted from the remaining non-noise subsequences. The second step is to cluster these extracted shapes. Although extracted from subsequences, these shapes form a non-overlapping set of time series subsequences and are hence amenable to meaningful clustering. The method is primarily designed for extracting and classifying shapes from very noisy real-world time series. Tests using artificial data with different levels of white noise and the red noise, and the real-world atmospheric turbulence data naturally characterised by strong red noise show that the method is able to correctly extract and cluster shapes from artificial data and that it has great potential for locating shapes in very noisy real-world time series.

The development of simple mathematical model for representing complicated real-life chemical reaction systems has been a fundamental issue in computational biology and bioinformatics. In particular, the accurate description of chemical events with multi-step chemical reactions has been regarded as an essential problem in chemistry and biophysics. To model chemical reaction systems in a manageable way, multi-step chemical reactions were normally simplified into a one-step reaction. In recent years, a number of modelling approaches have been attempted to use simplified model to describe multi-step chemical reactions accurately. In this work, we proposed a two-variable model to describe chemical events with multi-step chemical reactions. We introduced a new concept to represent the location of molecules in the multi-step reactions, and use it as the second indicator of the system dynamics. The accuracy of the proposed new model was evaluated via using a deterministic model. The proposed model has been applied to study the mRNA degradation process. Numerical simulations of the designed simplified models matched the simulations of multi-step chemical reactions very well.

This paper proposes and tests a methodology for selecting features and test cases with the goal of improving medium term bankruptcy prediction accuracy in large uncontrolled datasets of financial records. We propose a Genetic Programming and Neural Network based objective feature selection methodology to identify key inputs, and then use those inputs to combine multi-level Self-Organising Maps with Spectral Clustering to build clusters. Performing objective feature selection within each of those clusters, this research was able to increase out-of-sample classification accuracy from 71.3% and 69.8% on the Genetic Programming and Neural Network models respectively to 80.0% and 77.3%.

This paper proposes a new methodology to determine the relative performance of optimization algorithms across various classes of instances. Rather than reporting performance based on a chosen test set of benchmark instances, we aim to develop metrics for an algorithm's performance generalized across a diverse set of instances. Instances are summarized by a set of features that correlate with difficulty, and we propose methods for visualizing instances and algorithm performance in this high-dimensional feature space. The footprint of an algorithm is where good performance can be expected, and we propose new metrics to measure the relative size of an algorithm's footprint in instance space. The methodology is demonstrated using the Traveling Salesman Problem as a case study.

Identity crime is well known, prevalent, and costly; and credit application fraud is a specific case of identity crime. The existing nondata mining detection system of business rules and scorecards, and known fraud matching have limitations. To address these limitations and combat identity crime in real time, this paper proposes a new multilayered detection system complemented with two additional layers: communal detection (CD) and spike detection (SD). CD finds real social relationships to reduce the suspicion score, and is tamper resistant to synthetic social relationships. It is the whitelist-oriented approach on a fixed set of attributes. SD finds spikes in duplicates to increase the suspicion score, and is probe-resistant for attributes. It is the attribute-oriented approach on a variable-size set of attributes. Together, CD and SD can detect more types of attacks, better account for changing legal behavior, and remove the redundant attributes. Experiments were carried out on CD and SD with several million real credit applications. Results on the data support the hypothesis that successful credit application fraud patterns are sudden and exhibit sharp spikes in duplicates. Although this research is specific to credit application fraud detection, the concept of resilience, together with adaptivity and quality data discussed in the paper, are general to the design, implementation, and evaluation of all detection systems.

During the last decades, the disciplines of Data Mining and Operations Research have been working mostly independent of each other. However, the increasing complexity of today's applications in areas such as business, medicine, and science requires m

Multilinear subspace analysis (MSA) is a promising methodology for pattern-recognition problems due to its ability in decomposing the data formed from the interaction of multiple factors. The MSA requires a large training set, which is well organized in a single tensor, which consists of data samples with all possible combinations of the contributory factors. However, such a "complete" training set is difficult (or impossible) to obtain in many real applications. The missing-value problem is therefore crucial to the practicality of the MSA but has been hardly investigated up to present. To solve the problem, this paper proposes an algorithm named M(2)SA, which is advantageous in real applications due to the following: 1) it inherits the ability of the MSA to decompose the interlaced semantic factors; 2) it does not depend on any assumptions on the data distribution; and 3) it can deal with a high percentage of missing values. M(2)SA is evaluated by face image modeling on two typical multifactorial applications, i.e., face recognition and facial age estimation. Experimental results show the effectiveness of M(2) SA even when the majority of the values in the training tensor are missing.

One of the main difficulties in facial age estimation is the lack of sufficient training data for many ages. Fortunately, the faces at close ages look similar since aging is a slow and smooth process. Inspired by this observation, in this paper, instead of considering each face image as an example with one label (age), we regard each face image as an example associated with a label distribution. The label distribution covers a number of class labels, representing the degree that each label describes the example. Through this way, in addition to the real age, one face image can also contribute to the learning of its adjacent ages. We propose an algorithm named IIS-LLD for learning from the label distributions, which is an iterative optimization process based on the maximum entropy model. Experimental results show the advantages of IIS-LLD over the traditional learning methods based on single-labeled data.

The purpose of instance selection is to identify which instances (examples, patterns) in a large dataset should be selected as representatives of the entire dataset, without significant loss of information. When a machine learning method is applied to the reduced dataset, the accuracy of the model should not be significantly worse than if the same method were applied to the entire dataset. The reducibility of any dataset, and hence the success of instance selection methods, surely depends on the characteristics of the dataset, as well as the machine learning method. This paper adopts a meta-learning approach, via an empirical study of 112 classification datasets from the UCI Repository, to explore the relationship between data characteristics, machine learning methods, and the success of instance selection method.

An evolutionary approach for the optimization of microarray coatings produced via sol-gel chemistry is presented. The aim of the methodology is to face the challenging aspects of the problem: unknown objective function, high dimensional variable space, constraints on the independent variables, multiple responses, expensive or time-consuming experimental trials, expected complexity of the functional relationships between independent and response variables. The proposed approach iteratively selects a set of experiments by combining a multiob-jective Particle Swarm Optimization (PSO) and a multiresponse Multivariate Adaptive Regression Splines (MARS) model. At each iteration of the algorithm the selected experiments are implemented and evaluated, and the system response is used as a feedback for the selection of the new trials. The performance of the approach is measured in terms of improvements with respect to the best coating obtained changing one variable at a time (the method typically used by scientists). Relevant enhancements have been detected, and the proposed evolutionary approach is shown to be a useful methodology for process optimization with great promise for industrial applications.

Automatic estimation of human facial age is an interesting yet challenging topic appearing in recent years. Since different people might age in different ways, solving the problem of age estimation involves two semantic labels: identity and age. In this paper, aging face images are organized in a third-order tensor according to both identity and age. Due to the difficulty in data collection, the aging pattern for each person in the training set is always incomplete. Therefore, the tensor contains a large amount of missing values. Through a series of multilinear subspace analysis algorithms operating on tensor with missing values, the aging pattern contained in the training aging images can be iteratively learned and be used to predict the age of a given test image. In the experiment, the proposed method not only outperforms the existing algorithms, but also exceeds the human ability in age estimation.

The main difficulty in face image modeling is to decompose those semantic factors contributing to the formation of the face images, such as identity, illumination and pose. One promising way is to organize the face images in a higher-order tensor with each mode corresponding to one contributory factor. Then, a technique called Multilinear Subspace Analysis (MSA) is applied to decompose the tensor into the mode-$n$ product of several mode matrices, each of which represents one semantic factor. In practice, however, it is usually difficult to obtain such a complete training tensor since it requires a large amount of face images with all possible combinations of the states of the contributory factors. To solve the problem, this paper proposes a method named M$^2$SA, which can work on the training tensor with massive missing values. Thus M$^2$SA can be used to model face images even when there are only a small number of face images with limited variations which will cause missing values in the training tensor). Experiments on face recognition show that M$^2$SA can work reasonably well with up to $70\%$ missing values in the training tensor.

In this paper we propose a meta-learning inspired framework for analysing the performance of meta-heuristics for optimization problems, and developing insights into the relationships between search space characteristics of the problem instances and algorithm performance. Preliminary results based on several meta-heuristics for well-known instances of the Quadratic Assignment Problem are presented to illustrate the approach using both supervised and unsupervised learning methods.

Human age estimation by face images is an interesting yet challenging research topic emerging in recent years. This paper extends our previous work on facial age estimation (a linear method named AGES). In order to match the nonlinear nature of the human aging progress, a new algorithm named KAGES is proposed based on a nonlinear subspace trained on the aging patterns, which are defined as sequences of individual face images sorted in time order. Both the training and test (age estimation) processes of KAGES rely on a probabilistic model of KPCA. In the experimental results, the performance of KAGES is not only better than all the compared algorithms, but also better than the human observers in age estimation. The results are sensitive to parameter choice however, and future research challenges are identified.

Most work on multi-biometric fusion is based on static fusion rules which cannot respond to the changes of the environment and the individual users. This paper proposes adaptive multi-biometric fusion, which dynamically adjusts the fusion rules to suit the real-time external conditions. As a typical example, the adaptive fusion of gait and face in video is studied. Two factors that may affect the relationship between gait and face in the fusion are considered, i.e., the view angle and the subject-to-camera distance. Together they determine the way gait and face are fused at an arbitrary time. Experimental results show that the adaptive fusion performs significantly better than not only single biometric traits, but also those widely adopted static fusion rules including SUM, PRODUCT, MIN, and MAX.

“The authors would like to thank Y. Zhang, G. Li, and H. Dai for their help and Dr. A. Lanitis for providing the FG-NET Aging Database. Part of the work was done when Xin Geng was at the LAMDA Group, Nanjing University. Also, this research was partially supported by the National Science Foundation of China (60635030, 60325207) and the Foundation for the Author of National Excellent Doctoral Dissertation of China (200343).”

There usually exist many kinds of variations in face images taken under uncontrolled conditions, such as changes of pose, illumination, expression, etc. Most previous works on face recognition (FR) focus on particular variations and usually assume the absence of others. Instead of such a "divide and conquer" strategy, this paper attempts to directly address face recognition under uncontrolled conditions. The key is the individual stable space (ISS), which only expresses personal characteristics. A neural network named ISNN is proposed to map a raw face image into the ISS. After that, three ISS-based algorithms are designed for FR under uncontrolled conditions. There are no restrictions for the images fed into these algorithms. Moreover, unlike many other FR techniques, they do not require any extra training information, such as the view angle. These advantages make them practical to implement under uncontrolled conditions. The proposed algorithms are tested on three large face databases with vast variations and achieve superior performance compared with other 12 existing FR techniques.

While recognition of most facial variations, such as identity, expression and gender, has been extensively studied, automatic age estimation has rarely been explored. In contrast to other facial variations, aging variation presents several unique characteristics which make age estimation a challenging task. This paper proposes an automatic age estimation method named AGES (AGing pattErn Subspace). The basic idea is to model the aging pattern, which is defined as the sequence of a particular individual' s face images sorted in time order, by constructing a representative subspace. The proper aging pattern for a previously unseen face image is determined by the projection in the subspace that can reconstruct the face image with minimum reconstruction error, while the position of the face image in that aging pattern will then indicate its age. In the experiments, AGES and its variants are compared with the limited existing age estimation methods (WAS and AAS) and some well-established classification methods (kNN, BP, C4.5, and SVM). Moreover, a comparison with human perception ability on age is conducted. It is interesting to note that the performance of AGES is not only significantly better than that of all the other algorithms, but also comparable to that of the human observers.

Mining of text data from the Web has become a necessity in modern days due to the volumes of data available on the Web. While searching for information on the Web using search engines is popular, to analyze the content on large collections of Web pages, feature map techniques are still popular. One of the problems associated with processing large collections of text data from the Web using feature map techniques is the time taken to cluster them. This paper presents an algorithm based on a growing variant of the self organizing map called the HDGSOMr. This novel algorithm incorporates randomness into the self-organizing process to produce higher quality clusters within few epochs and utilizing smaller neighborhood sizes resulting in a significant reduction in overall processing time. Details of the HDGSOMr algorithm and results of processing large collections of text data proving the efficiency of the algorithm are also presented.

One of the major obstacles in using neural networks to solve combinatorial optimization problems is the convergence toward one of the many local minima instead of the global minima. In this letter, we propose a technique that enables a self-organizing neural network to escape from local minima by virtue of the intermittency phenomenon. It gives rise to novel search dynamics that allow the system to visit multiple global minima as meta-stable states. Numerical experiments performed suggest that the phenomenon is a combined effect of Kohonen-type competitive learning and the iterated softmax function operating near bifurcation. The resultant intermittent search exhibits fractal characteristics when the optimization performance is at its peak in the form of 1/f signals in the time evolution of the cost, as well as power law distributions in the meta-stable solution states. TheN-Queens problem is used as an example to illustrate the meta-stable convergence process that sequentially generates, in a single run, 92 solutions to the 8-Queens problem and 4024 solutions to the 17-Queens problem.

The growing self organizing map (GSOM) algorithm is a variant of the self organizing map (SOM). It has a dynamically growing structure that adapts to the natural structure of the data. It has been identified that the growing of the GSOM can get negatively affected when used with very large dimensional data such as those in text and DNA data sets. This paper addresses these issues and presents a modified version of the GSOM called the high dimensional GSOM (HDGSOM). The algorithm and experimental results showing the improved performance of the HDGSOM are also presented.

The self-organizing neural network (SONN) for solving general "0-1" combinatorial optimization problems (COPs) is studied in this paper, with the aim of overcoming existing limitations in convergence and solution quality. This is achieved by incorporating two main features: an efficient weight normalization process exhibiting bifurcation dynamics, and neurons with additive noise. The SONN is studied both theoretically and experimentally by using the N-queen problem as an example to demonstrate and explain the dependence of optimization performance on annealing schedules and other system parameters. An equilibrium model of the SONN with neuronal weight normalization is derived, which explains observed bands of high feasibility in the normalization parameter space in terms of bifurcation dynamics of the normalization process, and provides insights into the roles of different parameters in the optimization process. Under certain conditions, this dynamical systems view of the SONN reveals cascades of period-doubling bifurcations to chaos occurring in multidimensional space with the annealing temperature as the bifurcation parameter. A strange attractor in the two-dimensional (2-D) case is also presented. Furthermore, by adding random noise to the cost potentials of the network nodes, it is demonstrated that unwanted oscillations between symmetrical and "greedy" nodes can be sufficiently reduced, resulting in higher solution quality and feasibility.

Examines the use of neural networks as both a technique for pre-processing data and forecasting cash flow in the daily operations of a financial services company. The problem is to forecast the date when issued cheques will be presented by customers, so that the daily cash flow requirements can be forecast. These forecasts can then be used to ensure that appropriate levels of funds are kept in the company's bank account to avoid overdraft charges or unnecessary use of investment funds. The company currently employs an ad-hoc manual method for determining cash flow forecasts, and is keen to improve the accuracy of the forecasts. Unsupervised neural networks are used to cluster the cheques into more homogeneous groups prior to supervised neural networks being applied to arrive at a forecast for the date each cheque will be presented. Accuracy results are compared to the existing method of the company, together with regression and a heuristic method.

It has been over a decade since neural networks were first applied to solve combinatorial optimization problems. During this period, enthusiasm has been erratic as new approaches are developed and (sometimes years later) their limitations are realized. This article briefly summarizes the work that has been done and presents the current standing of neural networks for combinatorial optimization by considering each of the major classes of combinatorial optimization problems. Areas which have not yet been studied are identified for future research.

As an attempt to provide an organized way to study the chaotic structures and their effects in solving combinatorial optimization with chaotic neural networks (CNN's), a unifying framework is proposed to serve as a basis where the existing CNN models can be placed and compared. The key of this proposed framework is the introduction of an extra energy term into the computational energy of the Hopfield model, which takes on different forms for different CNN models, and modifies the original Hopfield energy landscape in various manners. Three CNN models, namely the Chen and Aihara model with self-feedback chaotic simulated annealing (CSA), the Wang and Smith model with timestep CSA, and the chaotic noise model, are chosen as examples to show how they can be classified and compared within the proposed framework.

After more than a decade of research, there now exist several neural-network techniques for solving NP-hard combinatorial optimization problems. Hopfield networks and self-organizing maps are the two main categories into which most of the approaches can be divided. Criticism of these approaches includes the tendency of the Hopfield network to produce infeasible solutions, and the lack of generalizability of the self-organizing approaches (being only applicable to Euclidean problems). This paper proposes two new techniques which have overcome these pitfalls: a Hopfield network which enables feasibility of the solutions to be ensured and improved solution quality through escape from local minima, and a self-organizing neural network which generalizes to solve a broad class of combinatorial optimization problems. Two sample practical optimization problems from Australian industry are then used to test the performances of the neural techniques against more traditional heuristic solutions.

Chen and Aihara recently proposed a chaotic simulated annealing approach to solving optimization problems. By adding a negative self-coupling to a network model proposed earlier by Aihara et al. and gradually removing this negative self-coupling, they used the transient chaos for searching and self-organizing, thereby achieving remarkable improvement over other neural-network approaches to optimization problems with or without simulated annealing. In this paper we suggest a new approach to chaotic simulated annealing with guaranteed convergence and minimization of the energy function by gradually reducing the time step in the Euler approximation of the differential equations that describe the continuous Hopfield neural network. This approach eliminates the need to carefully select other system parameters. We also generalize the convergence theorems of Chen and Aihara to arbitrarily increasing neuronal input-output functions and to less restrictive and yet more compact forms.

We examine the problem of assigning calls in a cellular mobile network to channels in the frequency domain. Such assignments must be made so that interference between calls is minimized, while demands for channels are satisfied. A new nonlinear integer programming representation of the static channel assignment (SCA) problem is formulated. We then propose two different neural networks for solving this problem. The first is an improved Hopfield (1982) neural network which resolves the issues of infeasibility and poor solution quality which have plagued the reputation of the Hopfield network. The second approach is a new self-organizing neural network which is able to solve the SCA problem and many other practical optimization problems due to its generalizing ability. A variety of test problems are used to compare the performance of the neural techniques against more traditional heuristic approaches. Finally, extensions to the dynamic channel assignment problem are considered.

In this paper, a distinction is drawn between research which assesses the suitability of the Hopfield network for solving the travelling salesman problem (TSP) and research which attempts to determine the effectiveness of the Hopfield network as an optimization technique. It is argued that the TSP is generally misused as a benchmark for the latter goal, with the existence of an alternative linear formulation giving rise to unreasonable comparisons.