Program
Invited Conference
Gabriela Ochoa
The Cartography of Computational Search Spaces. Friday, October 27, 11AM.
Abstract: This talk will present our recent findings and visual (static and animated) maps characterising combinatorial and program search spaces. We seek to lay the foundations for a new perspective to understand problem structure and improve heuristic search algorithms: search space cartography. A multitude of heuristic and bio-inspired search algorithms has been proposed, each trying to be more powerful and innovative. However, little attention has been devoted to understanding the problems’ structure and what makes them hard to solve for a given algorithm. Formal theoretical results are difficult to obtain, and they may only apply to problem classes and algorithms chosen more for their amenability to analysis than for their relevance and difficulty. Heuristic methods operate by searching a large space of candidate solutions. The search space can be regarded as a spatial structure where each point (candidate solution) has a height (objective or fitness function value) forming a fitness landscape surface. The performance of optimisation algorithms crucially depends on the fitness landscape structure, and the study of landscapes offers an alternative to problem understanding where realistic formulations and algorithms can be analysed. Most fitness landscapes analysis techniques study the local structure of search spaces. There is currently a lack of tools to study instead their global structure, which is known to impact algorithms’ performance. Our recently proposed model, local optima networks, fills this gap by bringing tools from complex networks to study optimisation. This model provides fundamental new insight into the structural organisation and the connectivity pattern of a search space with given move operators. Most importantly, it allows us to visualise realistic search spaces in ways not previously possible and brings a whole new set of quantitative network metrics for characterising them.
Short Bio: Gabriela Ochoa has worked on Evolutionary Computation since the late 90s. Her very first article used genetic algorithms to evolve plant-like structures encoded as Lindenmayer systems. During her PhD at the University of Sussex, UK she worked with the notion of error thresholds in genetic algorithms. She held academic positions at the University Simon Bolivar in Caracas, Venezuela and the University of Nottingham, UK before joining the University of Stirling, Scotland in 2012. Her current research includes autonomous search and hyper-heuristics, fitness landscape analysis and applications to combinatorial optimisation, software engineering and healthcare. She is an associate editor of the Journal of Evolutionary Computation, MIT Press and the IEEE Transactions of Evolutionary Computation. She has served as an organiser for EvoCOP, FOGA and PPSN, is a member of the board of ACM SIGEVO, and served as the Editor-in-Chief for GECCO 2017.
Jean-Daniel Fekete
Progressive Data Analysis: a new computation paradigm for scalability in exploratory data analysis. Wednesday, October 25, 2PM.
Abstract: Exploring data requires a short feedback loop, with a latency of at most 10 seconds because of human cognitive capabilities and limitations. When data becomes large or analyses become complex, sequential computations can no longer be completed in a few seconds and interactive exploration is severely hampered. This talk will describe a novel computation paradigm called Progressive Data Analysis that brings at the programming language level the low-latency guarantee by performing computations in a progressive fashion. Moving this progressive computation at the language level relieves the programmer of exploratory data analysis systems from implementing the whole analytics pipeline in a progressive way from scratch, streamlining the implementation of scalable exploratory analytics systems. I will describe the new paradigm, report on novel experiments showing that human can cope effectively with progressive systems, and show demos using a prototype implementation called ProgressiVis, explain the requirements it implies through exemplar applications, and present opportunities and challenges ahead, in the domains of visualization and machine-learning.
List of papers
List of papers with status "Definitely accept as talk"
• Paper #3 - Reinforcement Learning as a model of aposematism
Eduardo Alonso, Mark Broom, Jan Teichmann
• Paper #5 - Evolutionary learning of fire fighting strategies
Martin Kretschmer, Elmar Langetepe
• Paper #6 - Semantics-based Crossover for Program Synthesis in Genetic Programming
Stefan Forstenlechner, David Fagan, Miguel Nicolau, Michael O'Neill
• Paper #7 - On the Design of a Master-Worker Adaptive Algorithm Selection Framework
Christopher Jankee, Sébastien Verel, Bilel Derbel, Cyril Fonlupt
• Paper #8 - Evolutionary Optimization of Tone Mapped Image Quality Index
Xihe Gao, Stephen Brooks, Dirk V. Arnold
• Paper #16 - On the Use of Dynamic GP Fitness Cases in Static and Dynamic Optimisation Problems
Edgar Galvan-Lopez, Lucia Vazquez-Mendoza, Marc Schoenauer, Leonardo Trujillo
• Paper #17 - H-ACO: A Heterogeneous Ant Colony Optimisation approach with Application to the Travelling Salesman Problem
Ahamed Fayeez Tuani Ibrahim, Edward Keedwell, Matthew Collett
• Paper #18 - LIDeOGraM : An interactive evolutionary modelling tool.
Thomas Chabin, Marc Barnabé, Nadia Boukhelifa, Fernanda Fonseca, Alberto Tonda, Hélène Velly, Benjamin Lemaitre, Nathalie Perrot, Evelyne Lutton
• Paper #19 - Offline Learning for Selection Hyper-heuristics with Elman Networks
William Yates, Edward Keedwell
• Paper #26 - A New High-Level Relay Hybrid Metaheuristic for Black-Box Optimization Problems
Julien Lepagnot, Lhassane Idoumghar, Mathieu Brévilliers
• Paper #27 - Distance Cooperation between Hybrid Iterative Tabu Search
Omar ABDELKAFI, Lhassane IDOUMGHAR, Julien LEPAGNOT
• Paper #28 - Sampled Walk and Binary Fitness Landscapes Exploration
Sara Tari, Matthieu Basseur, Adrien Goëffon
• Paper #30 - A fitness landscape view on the tuning of an asynchronous master-worker EA for nuclear reactor design
Mathieu Muniglia, Sébastien Verel, Jean-Charles Le Pallec, Jean-Michel Do
• Paper #33 - Comparison of Acceptance Criteria in Randomized Local Searches
Alberto Franzin, Thomas Stützle
• Paper #34 - Automatic configuration of GCC using irace
Leslie Pérez Cáceres, Federico Pagnozzi, Alberto Franzin, Thomas Stützle
• Paper #36 - MEMSA: a robust Parisian EA for multidimensional multiple sequence alignment
Julie Thompson, Pierre Collet
• Paper #37 - Basic, Dual, Adaptive, and Directed Mutation Operators in the Fly Algorithm
Zainab Ali Abbood, Franck Vidal
List of papers with status "Definitely accept as poster"
• Paper #4 - Reinforcement learning is an effective strategy to create phenotypic variation and a potential mechanism for the initial evolution of learning
Eduardo Alonso, Mark Broom, Jan Teichmann
• Paper #9 - Learning new Term Weighting Schemes with Genetic Programming
Ahmad Mazyad, Fabien Teytaud, Cyril Fonlupt
• Paper #10 - The Ant Reconciliation Algorithm (ARA): Ant-hill learning for label matching
Pierre Parrend, Camille Maller, Etienne Dietrich
• Paper #11 - An improved particle swarm optimization algorithm for MRI image segmentation
Thuy Xuan Pham, Patrick Siarry, Hamouche Oulhadj
• Paper #12 - Crowd-Sourced Optimisation of Procedural Animation Systems
Gareth Henshall, William Teahan, Llyr ap Cenydd
• Paper #21 - Optimization of allelic combinations controlling parameters of a peach quality model
Bénédicte Quilot-Turion, Michel Génard, Pierre Valsesia, Mohamed-mahmoud Memmah
• Paper #35 - Towards a More Practice-Aware Runtime Analysis
Eduardo Carvalho Pinto, Carola Doerr