Vineet R. Khare

Performance scaling of multi-objective evolutionary algorithms

MOEAs are getting immense popularity in the recent past, mainly because of their ability to find a wide spread of Pareto-optimal solutions in a single simulation run. Various evolutionary approaches to multi-objective optimization have been proposed since 1985. Some of fairly recent ones are NSGA-II, SPEA2, PESA (which are included in this study) and others. They all have been mainly applied to two to three objectives. In order to establish their superiority over classical methods and demonstrate their abilities for convergence and maintenance of diversity, they need to be tested on higher number of objectives. In this study, these state-of-the-art MOEAs have been investigated for their scalability with respect to the number of objectives (2 to 8). They have also been compared on the basis of -(1) Their ability to converge to Pareto front, (2) Diversity of obtained non-dominated solutions and (3) Their running time. Four scalable test problems (DTLZ1, 2, 3 and 6) are used for the comparative study.

Co-evolutionary modular neural networks for automatic problem decomposition

Decomposing a complex computational problem into sub-problems, which are computationally simpler to solve individually and which can be combined to produce a solution to the full problem, can efficiently lead to compact and general solutions. Modular neural networks represent one of the ways in which this divide-and-conquer strategy can be implemented. Here we present a co-evolutionary model which is used to design and optimize modular neural networks with task-specific modules. The model consists of two populations. The first population consists of a pool of modules and the second population synthesizes complete systems by drawing elements from the pool of modules. Modules represent a part of the solution, which co-operates with others in the module population to form a complete solution. With the help of two artificial supervised learning tasks created by mixing two sub-tasks we demonstrate that if a particular task decomposition is better in terms of performance on the overall task, it can be evolved using this co-evolutionary model.

A fuzzy logic based approach for modeling quality and reliability related customer satisfaction in the automotive domain

This paper presents an approach to assess quality and reliability related customer satisfaction from field failure data at each individual customer level. The quality satisfaction has been modeled based on number of failures and severity of failures, while, reliability satisfaction has been modeled based on number of visits to dealer and time span between visits. The satisfaction modeled at an individual vehicle (customer) level is further aggregated to a vehicle model level to determine overall satisfaction of customers with that specific vehicle model. A fuzzy logic approach is used to construct the satisfaction model. A grid search technique is used to tune the model parameters such that the output of the model for specific vehicle models matches with survey based ratings assigned to the vehicle models.

Credit Assignment Among Neurons in Co-evolving Populations

Different credit assignment strategies are investigated in a two level co-evolutionary model which involves a population of Gaussian neurons and a population of radial basis function networks consisting of neurons from the neuron population. Each individual in neuron population can contribute to one or more networks in network population, so there is a two-fold difficulty in evaluating the effectiveness (or fitness) of a neuron. Firstly, since each neuron only represents a partial solution to the problem, it needs to be assigned some credit for the complete problem solving activity. Secondly, these credits need to be accumulated from different networks the neuron participates in. This model, along with various credit assignment strategies, is tested on a classification (Heart disease diagnosis problem from UCI machine learning repository) and a regression problem (Mackey-Glass time series prediction problem).

Multi-network evolutionary systems and automatic decomposition of complex problems

Multi-network systems, i.e. multiple neural network systems, can often solve complex problems more effectively than their monolithic counterparts. Modular neural networks (MNNs) tackle a complex problem by decomposing it into simpler subproblems and then solving them. Unlike the decomposition in MNNs, a neural network ensemble usually includes redundant component nets and is often inspired by statistical theories. This paper presents different types of problem decompositions and discusses the suitability of various multi-network systems for different decompositions. A classification of various multi-network systems, in the context of problem decomposition, is obtained by exploiting these differences. Then a specific type of problem decomposition, which gives no information about the subproblems and is often ignored in literature, is discussed in detail and a novel MNN architecture for problem decomposition is presented. Finally, a co-evolutionary model is presented, which is used to design and optimize such MNNs with subtask specific modules. The model consists of two populations. The first population consists of a pool of modules and the second population synthesizes complete systems by drawing elements from the pool of modules. Modules represent a part of the solution, which co-operate with each other to form a complete solution. Using two artificial supervised learning tasks, constructed from smaller subtasks, it can be shown that if a particular task decomposition is better than others, in terms of performance on the overall task, it can be evolved using the co-evolutionary model.

Development, analysis and applications of a quantitative methodology for assessing customer satisfaction using evolutionary optimization

Graphical abstractDisplay Omitted HighlightsQuantitative modeling of customer satisfaction for consumer vehicles is proposed.Real-world service and sales datasets of five vehicle models are used.Model sensitivity to various features of the service datasets is studied.Classification rules for identifying dissatisfied customers are obtained.Method for identifying high-priority vehicular problems is proposed. Consumer-oriented companies are getting increasingly more sensitive about customers perception of their products, not only to get a feedback on their popularity, but also to improve the quality and service through a better understanding of design issues for further development. However, a consumers perception is often qualitative and is achieved through third party surveys or the companys recording of after-sale feedback through explicit surveys or warranty based commitments. In this paper, we consider an automobile companys warranty records for different vehicle models and suggest a data mining procedure to assign a customer satisfaction index (CSI) to each vehicle model based on the perceived notion of the level of satisfaction of customers. Based on the developed CSI function, customers are then divided into satisfied and dissatisfied customer groups. The warranty data are then clustered separately for each group and analyzed to find possible causes (field failures) and their relative effects on customers satisfaction (or dissatisfaction) for a vehicle model. Finally, speculative introspection has been made to identify the amount of improvement in CSI that can be achieved by the reduction of some critical field failures through better design practices. Thus, this paper shows how warranty data from customers can be utilized to have a better perception of ranking of a product compared to its competitors in the market and also to identify possible causes for making some customers dissatisfied and eventually to help percolate these issues at the design level. This closes the design cycle loop in which after a design is converted into a product, its perceived level of satisfaction by customers can also provide valuable information to help make the design better in an iterative manner. The proposed methodology is generic and novel, and can be applied to other consumer products as well.

Pattern classification driven enhancements for human-in-the-loop decision support systems

Data mining has been a key technology in the warranty sector for mass manufacturers to understand and improve product quality, reliability and durability. Cost savings is an important aspect of business which calls for processes that are error proof. Pattern classification methods applied to the diagnostic data could help build error proof processes by improving the diagnostic technology. In this paper we present a case study from the automotive warranty and service domain involving a human-in-the-loop decision support system (HIL-DSS). The automotive manufacturers offer warranties on products, made of parts from different suppliers, and rely on a dealer network to assess warranty claims. The dealers use diagnostic equipment manufactured by third parties and also draw on their own expertise. In addition, a subject matter expert (SME) assesses these collective decisions to distinguish between inaccurate diagnoses by the dealers or an inadequate decision algorithm in the diagnostic equipment. Altogether this makes a comprehensive HIL-DSS. The proposed methodology continuously learns from collective decision making systems, enhances the diagnostic equipment, adds to the knowledge of dealers and minimizes the SME involvement in the review process of the overall system. Improving the diagnostic equipment helps in better warranty servicing, whereas improvements in the human expert knowledge help prevent field error and avoid customer dissatisfaction due to improper fault diagnosis.

Environments conducive to evolution of modularity

Modularity has been recognised as one of the crucial aspects of natural complex systems. Since these are results of evolution, it has been argued that modular systems must have selective advantages over their monolithic counterparts. Simulation results with artificial neuro-evolutionary complex systems, however, are indecisive in this regard. It has been shown that advantages of modularity, if judged on a static task, in these systems are very much dependent on various factors involved in the training of these systems. We present a couple of dynamic environments and argue that environments like these might be partly responsible for the evolution of modular systems. These environments allow for a better, more direct use of structural information present within modular systems hence limit the influence of other factors. We support these arguments with the help of a co-evolutionary model and a fitness measure based on system performance in these dynamic environments.

Automotive field failure analysis based on mileage — Feasibility & benefits

Most failures in the automotive systems depend on age and accumulated usage. Typically, these systems are covered under warranty for months-in-service (i.e. MIS/age) and a certain amount of usage (mileage) after the sales of the products. Warranty analysis of these systems enables manufacturers to understand field failures, and identify focus areas to make product improvements. Typically warranty analysis is performed based on MIS. However, mileage based warranty analysis has two added benefits - (1) some failures are, by their physical nature, related to mileage rather than age. Hence, mileage is a better indicator for observing and quantifying these failures. (2) Our observations also indicate that most vehicles leave warranty due to the mileage restrictions to the warranty coverage, rather than MIS. In such a scenario, mileage-based warranty calculations can provide us with early information. Warranty analysis based on mileage has been presented in the literature as a supplement to the traditional MIS based analysis. However, these are based on the following two major assumptions, and their validity has not been established yet: · The accumulation of miles is approximately linear with age. · The distribution of mileage accumulation rate in vehicles without any claims is same as that of vehicles which have at least one warranty claim. Using the real-time diagnostic data collected from Telematics systems, this work demonstrates that the above two assumptions are valid. As a result, we can achieve accurate warranty rates and take the advantage of early detection of problems using mileage. Early detection is based on high usage vehicles. We also demonstrate that the proportion of such vehicles is statistically significant, which enables mileage based analysis feasible earlier than age based analysis. This lead time is of crucial importance to the OEMs, especially close to the launch of new products, where they can identify and rectify field failures early.

Identification and impact assessment of high-priority field failures in passenger vehicles using evolutionary optimization

This paper presents a method for prioritizing field failures in passenger vehicles based on their potential for improvement in the Customer Satisfaction Index (\({\text{ CSI}}_{QSR}\)). \({\text{ CSI}}_{QSR}\) refers to Customer Satisfaction Index pertaining to quality, service and reliability of the vehicle and is referred to as simply ‘CSI’ in this paper. A novel method for quantitative modeling of the CSI function using an evolutionary approach was presented in [3]. Such a CSI function can be used to capture individual customer’s perception of a vehicle model as well as to compare overall CSI of multiple vehicle models. This work is firstly aimed at improving the previous modeling technique and validating it against Consumer Reports reliability ratings. More importantly, it presents a procedure for identifying high impact field failures based on their CSI Improvement Potential (CIP). These high priority field failures can then be further studied for root cause analysis.