Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

https://www.cirrelt.ca/DocumentsTravail/CIRRELT-2011-62.pdf

A review of dynamic vehicle routing problems

This paper surveys the research on the metaheuristics for the Vehicle Routing Problem with Time Windows (VRPTW). The VRPTW can be described as the problem of designing least cost routes from one depot to a set of geographically scattered points. The routes must be designed in such a way that each point is visited only once by exactly one vehicle within a given time interval; all routes start and end at the depot, and the total demands of all points on one particular route must not exceed the capacity of the vehicle. Metaheuristics are general solution procedures that explore the solution space to identify good solutions and often embed some of the standard route construction and improvement heuristics described in the first part of this article. In addition to describing basic features of each method, experimental results for Solomon's benchmark test problems are presented and analyzed.

/pdf/vehicle-routing-problem-with-time-windows-part-ii-4tzmnpdxqc.pdf

Vehicle Routing Problem with Time Windows, Part II: Metaheuristics

Since the late 70s, much research activity has taken place on the class of dynamic vehicle routing problems DVRP, with the time period after year 2000 witnessing a real explosion in related papers. Our paper sheds more light into work in this area over more than 3 decades by developing a taxonomy of DVRP papers according to 11 criteria. These are 1 type of problem, 2 logistical context, 3 transportation mode, 4 objective function, 5 fleet size, 6 time constraints, 7 vehicle capacity constraints, 8 the ability to reject customers, 9 the nature of the dynamic element, 10 the nature of the stochasticity if any, and 11 the solution method. We comment on technological vis-i-vis methodological advances for this class of problems and suggest directions for further research. The latter include alternative objective functions, vehicle speed as decision variable, more explicit linkages of methodology to technological advances and analysis of worst case or average case performance of heuristics. © 2015 Wiley Periodicals, Inc. NETWORKS, Vol. 671, 3-31 2016

/pdf/dynamic-vehicle-routing-problems-three-decades-and-counting-3phqfejokj.pdf

Dynamic vehicle routing problems: Three decades and counting

The emergence of metaheuristics for solving difficult combinatorial optimization problems is one of the most notable achievements of the last two decades in operations research This paper provides an account of the most recent developments in the field and identifies some common issues and trends Examples of applications are also reported for vehicle routing and scheduling problems

Metaheuristics in Combinatorial Optimization

https://www.iro.umontreal.ca/~potvin/ying.pdf

Vehicle routing and scheduling with dynamic travel times

Vehicle dispatching consists of allocating real-time service requests to a fleet of moving vehicles. In this paper, each vehicle is associated with a vector of attribute values that describes its current situation with respect to new incoming service requests. Using this attribute description, a utility function aimed at approximating the decision process of a professional dispatcher is constructed through genetic programming. Computational results are reported on requests collected from a courier service company and a comparison is provided with a neural network model and a simple dispatching policy.

Decision support for vehicle dispatching using genetic programming

Wireless sensor networks WSNs have boomed in this last decade. They are involved in all aspects of our daily lives and make it easier. Despite the great strides in these networks, several problems arose and are still open. The localisation of sensor networks nodes is one of the most challenging problem that has attracted many researches. Consequently, a lot of work has been carried out to provide accurate location information of sensors and sensing field. In this paper, we present a taxonomy of the most important localisation technologies proposed for WSNs. We survey different published localisation algorithms in the literature while giving an up-to-date with the most recent schemes. Then we highlight their characteristics and describe metrics used for their classification. A complete comparison will be given and also directions of future research will be discussed.

Localisation algorithms for wireless sensor networks: a review

The routing in communication networks is typically a multicriteria decision making (MCDM) problem. However, setting the parameters of most used MCDM methods to fit the preferences of a decision maker is often a difficult task. A Russian doll method able to choose the best multicriteria solution according to a context defined beforehand is proposed. This context is given by a set of nested boxes in the criteria space, the shapes of which can be established from objective facts such as technical standards, technical specifications, etc. This kind of method is well suited for self-adaptive systems because it is designed to be able to give pertinent results without interaction with a decision maker, whatever the Pareto front. The Russian doll multicriteria decision method is used with a reinforcement learning to optimize the routing in a mobile ad-hoc network. The results on a case study show that the routing can be finely controlled because of the possibility to include as much parameters as desired to adjust the search of the best solution on Pareto fronts a priori unknown. These results are clearly better than those obtained with the optimization of a weighted sum or the minimization of a Chebyshev distance to a reference point.

/pdf/multicriteria-reinforcement-learning-based-on-a-russian-doll-3aoctecbgg.pdf

Multicriteria reinforcement learning based on a Russian doll method for network routing

The vehicle routing problem with time windows is used to model. many realistic applications and is the focus of very intensive research. In this paper, we generalize two route construction heuristics for this problem. The additional parameter values of the new heuristics are optimized with a genetic algorithm. Numsrical results on a standard set of problems are also reported.

Ilham Benyahia

Papers

Vehicle routing and scheduling with dynamic travel times

Decision support for vehicle dispatching using genetic programming

Localisation algorithms for wireless sensor networks: a review

Multicriteria reinforcement learning based on a Russian doll method for network routing

Generalization and refinement of route construction heuristics using genetic algorithms