Plenary Speakers:
-
Mauricio G.C. Resende (AT&T Labs Research, Florham Park, New Jersey, USA)
Hybridization of heuristics with biased random-key genetic algorithms: Application to 2-dim orthogonal packing and 3-dim bin packingJoint work José F. Gonçalves of U. do Porto (Portugal)
A biased random-key genetic algorithm (BRKGA) is a general search metaheuristic for finding optimal or near-optimal solutions to hard combinatorial optimization problem (Gonçalves & Resende, 2011). It is derived from the random-key genetic algorithm of Bean (1994), differing in the way solutions are combined to produce offspring. BRKGAs have three key features that specialize genetic algorithms:
- A fixed chromosome encoding using a vector of n random keys or alleles over the interval [0,1), where the value of n depends on the instance of the optimization problem;
- A well-defined evolutionary process adopting parameterized uniform crossover (Spears & DeJong, 1991) to generate offspring and thus evolve the population;
- The introduction of new chromosomes called mutants in place of the mutation operator usually found in genetic algorithms.
Such features simplify and standardize the metaheuristic with a set of self-contained tasks from which only chromosome decoding is problem-dependent. Decoding constructs a solution to the underlying optimization problem, from which the objective function value or fitness can be computed. BRKGAs allow for natural hybridizations of heuristics.
In this talk, we introduce BRKGAs and describe hybrid heuristics based on BRKGAs for orthogonal 2- and 3-dimensional packing as well as for 2- and 3-dimensional bin packing. These heuristics have produced new best-known solutions for a number of benchmark instances from the packing literature.
-
Roberto Battiti (University of Trento, Italy)
Learning and Intelligent Optimization: one ring to rule them allAlmost by definition, optimization is a source of a tremendous power for automatically improving processes, decisions, products and services. But its potential power is still largely unexploited in most real-world contexts. We argue that one of the main reasons blocking its widespread adoption is that standard optimization assumes the existence of a function f(x) to be minimized, while in most real-world business contexts this function does not exist or is extremely difficult and costly to build by hand. Machine learning comes to the rescue: the function (the model) can be built by machine-learning from abundant data. By Learning and Intelligent Optimization (LION) we mean this combination of learning from data and optimization which can be applied to complex, dynamic, stochastic contexts. The automation level will be increased; more power will be directly in the hands of decision-makers without resorting to intermediate layers of data scientists. Reaching this goal is a huge challenge and it will require research at the boundary between two areas, machine learning and optimization, which have been traditionally separated.