European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning

Call for papers
Special sessions
Author guidelines
Infos speakers
Social events
Get a password

Login:  Password:   Password forgotten ?

Special sessions

Special sessions are organized by renowned scientists in their respective fields. Papers submitted to these sessions are reviewed according to the same rules as any other submission. Authors who submit papers to one of these sessions are invited to mention it on the author submission form; submissions to the special sessions must follow the same format, instructions and deadlines as any other submission, and must be sent according to the same procedure.

The following special sessions will be organized at ESANN2017:

Environmental signal processing: new trends and applications
Organized by Gilles Delmaire, Matthieu Puigt, Gilles Roussel (Université du Littoral Côte d'Opale, France)

In the last years, environmental monitoring was shown to be a major application field of modern signal processing/machine learning techniques. Indeed, in addition to the natural interest of such interdisciplinary topic, it may provide some interesting problems for which ad hoc methods, e.g., matrix/tensor factorization, matrix/tensor completion, blind sensor calibration, or source separation, must be proposed. The scope of this special session will aim to put on light both specific environmental problems and newly designed signal processing methods to tackle them.

Possible topics are related but not only to the following aspects:

  • matrix/tensor factorization (including source separation, deconvolution, non-negative matrix/tensor factorization, dictionary learning, sparse approximation, random matrix theory, etc);
  • matrix/tensor completion;
  • blind sensor calibration;
  • chemical sensor processing;
  • compressed sensing.

Biomedical data analysis in translational research: integration of expert knowledge and interpretable models
Organized by Gyan Bhanot (Rutgers University, New Jersey, USA), Michael Biehl (University of Groningen, The Netherlands), Thomas Villmann (Univ. of Applied Sciences Mittweida, Germany), Dietlind Zühlke (Seven Principles, Germany)

New technologies in various fields of biomedical research have led to a dramatic increase of the amount of electronic data that is available. Not only is the number of patients or amount of disease specific data increasing, but so is the structural complexity of the data, in terms of its dimensionality, multi-modality and inhomogeneity.

A significant problem, recognized by both the bio-medical and computational community, is the lack of coordination among researchers in these disparate communities. On the one hand, integration of expert knowledge is instrumental for successful data analysis and modelling. On the other hand, methods and models should be transparent and interpretable in order to facilitate fruitful trans-disciplinary collaboration.

This special session is meant to attract researchers who develop, investigate, or apply methods of machine learning and statistics in biomedical data analysis, experts from knowledge representation and integration as well as bio-medical researchers with a strong interest in computation and interpretable models.

Topics include, but are not restricted to:

  • Structured, inhomogeneous and multi-modal biomedical data
  • Feature selection and identification of biomarkers
  • Interpretable systems for diagnosis and classification
  • Generative models of bio-medical processes
  • Visual analytics and data mining
  • Big data mining for clinical impact

Processing, Mining and Visualizing Massive Urban Data
Organized by Etienne Côme (UPE-Ifsttar, France), Pierre Borgnat (ENS Lyon, France), Latifa Oukhellou (UPE-Ifsttar, France)

The development of smart technologies, as well as the advent of new observation capabilities, has boosted the availability of massive urban datasets that can greatly benefit urban studies. Indeed, a large amount of data are collected by various urban sensors including smart meters data, GSM, Wi-Fi or Bluetooth records, ticketing data, geo-tagged posts on social networks, etc. The analysis of such digital records can help to build decision-making tools (for analytical, forecasting and display purposes) with a view to better understand the citizen behaviors, to give urban stakeholders the capacity to achieve a better planning and scale-up infrastructures and to provide better services to citizens in order to foster the development of the city and the quality of life. The purpose of this special session is to bring together recent research studies aiming at mining and visualizing massive urban data. Contributions focusing on advanced data mining and machine learning approaches as well as visualization methods dedicated to exploiting such datasets are encouraged. Application domains can include Energy, Mobility and Smart transportation, Health, ….

Randomized Machine Learning approaches: analysis and developments
Organized by Claudio Gallicchio (University of Pisa, Italy), José D. Martín-Guerrero (University of Valencia, Spain), Alessio Micheli (University of Pisa, Italy), Emilio Soria (University of Valencia, Spain)

Randomness has always been present in one or other form in Machine Learning (ML) models; for instance, data sets have been randomly split into two training and test sets; also, random initializations of the parameters have always been common, and even advisable. However, the last few years have observed a change of paradigm, in which randomness is not only accessory, but plays a key role in many occasions, e.g., in the well-known random forests. In the Neural Network (NN) area, since its origins, randomness gave rise to a rich set of models, which have been recently exploited especially for efficiency aims. However, the bias induced by the use NN with random weights deserves further analysis, especially in the novel advances in the fields of deep NN, dynamical systems (Recurrent NN), and NN for learning in structured domains.

This session calls for contributions dealing with new analyses and developments of randomized approaches for ML, as a way of enhancing their understanding and performance. The session is also open to critical analysis of randomized approaches and to works that point out potential flaws and limitations of randomized machine learning models.

The topics of the session include, but are not limited, to the following:

  • Neural Networks with random weights
  • Extreme Learning Machines, Random Vector Functional-Link Networks
  • Reservoir Computing
  • Deep Randomized Neural Networks
  • Random learning algorithms
  • Random ensembles: random forests, extremely randomized trees, random combinations of neural networks, etc.
  • Novel randomized models for Structured Data (sequences, trees, graphs)
  • Random Projections
  • Randomized search of optimal parameters
  • Efficient design of random models for Big Data
  • Theory of Randomized Neural Networks
  • Open issues and limitations: learnability, range of applicability, stability and efficiency, comparisons
  • Biological plausibility/inspiration of Randomized Neural Networks
  • Parallel Computing for Randomized models
  • Linear basis expansion and Kernel approaches
  • Bayesian approaches
  • Development of new ML models using random structures
  • Performance assessment
  • Applications

Deep and kernel methods: best of two worlds
Organized by Lluís A. Belanche, Marta R. Costa-jussà (Universitat Politècnica de Catalunya, Barcelona, Spain)

Multilayer neural networks have experienced a rebirth in the data analysis field, displaying impressive results, extending even to classical artificial intelligence domains, such as game playing, computer vision, natural language and speech processing. The versatility of such methods have lead deep (semi)-parametric models to get over well-established learning methods, like kernel machines or classical statistical techniques. However, their training is a delicate and costly optimization problem that raises many practical challenges. On the other hand, kernel methods usually involve solving a tractable convex problem and are able to handle non-vectorial data directly, leading to a higher expressive power. Their main drawback is arguably their complexity being dependent on the number of data points, both at training and model evaluation times. A natural and emerging field of research is given by their hybridization, which can done in many fruitful ways. Many ideas from the deep learning field can be transferred to the kernel framework and viceversa.

This special session aims at all aspects of deep architectures, be theoretical or methodological developments, comparative analyses, or applications. A special emphasis is given to new ideas to bridge the gap between the fields of deep and kernel learning, as well as the understanding of their respective weak and strong points.

The topics of the session include, but are not limited to,

  • Applications of deep architectures in data representation and analysis, including structured or non-vectorial inputs or outputs
  • Natural language and speech processing; machine translation and language modeling; reordering, tuning and rescoring
  • Scalability/efficiency of deep neural networks and large-scale kernel machines
  • Heterogeneous data and meta-data; structured relationships among data
  • Applications in neuroscience, computer vision, (bio)acoustic signals and mechanisms
  • Statistical or stability analysis, visualization of learning, generalization bounds
  • Novel deep(er) architectures/algorithms for data representation and learning (using kernels or not)
  • Recursive and iterative kernels and their relation to deep neural architectures
  • Emulation of multilayer machines by shallow architectures and vice versa
  • Randomized (approximate) feature maps to scale-up kernel methods
  • Derivation of efficient layer-by-layer algorithms for training such networks; reductions in the computational complexity
  • Comparisons of deep architectures to shallow architectures

Algorithmic Challenges in Big Data Analytics
Organized by: Veronica Bolon-Canedo, Amparo Alonso-Betanzos (University of A Coruña, Spain), Beatriz Remeseiro (University of Barcelona, Spain), David Martinez-Rego (University College London, UK), Konstantinos Sechidis (University of Manchester, UK)

In the past few years, the advent of Big Data has brought unprecedented challenges to machine learning researchers. Dealing with huge volumes of data, both in terms of instances and features, makes the learning task more complex and computationally demanding than ever.

Processing these massive datasets is key to providing a wealth of information, but at the same time is a challenge for machine learning researchers, who see how classic algorithms are now useless. The community expects new methods that not only allow accurate analysis of the available data, but which are also robust and scalable when dataset sizes increase. In other words, the challenge now is to find “good enough” solutions as “fast” as possible and as “efficiently” as possible. This issue becomes critical in situations in which there exist temporal or spatial constraints like real-time applications or unapproachable computational problems requiring learning.

We invite papers aiming to examine the recent progress in the field, together with new open challenges derived from the increased data availability. In particular, topics of interest include, but are not limited to:

  • Pre-processing, processing and post-processing of Big Data.
  • Methods, algorithms and theory for Big Data analytics.
  • Recent advances and challenges in machine learning for Big Data.
  • Distributed learning in the context of Big Data.
  • Deep learning with massive-scale datasets.
  • Applications: healthcare, social media, bioinformatics, genomics, finance, surveillance, etc.
Submitted papers will be reviewed according to the ESANN reviewing process and will be evaluated on their scientific value: originality, correctness, and writing style.

Machine Learning in Biomorphic Robots
Organized by Nigel Crook, Matthias Rolf, Tjeerd olde Scheper (Oxford Brookes Univ., UK)

Biological organisms are a primary inspiration for the design, and development of robotic devices. This informs not only the physical design of the robot, but also its movement and sensing characteristics. Biologically inspired robots have often novel physical designs that challenge standard approaches to control. The special session Machine Learning in Biomorphic Robots seeks to bring together state-of-the-art research on the application of machine learning to all aspects of biomorphic robot development, dynamics and control. The session will include a tutorial session on bio-inspired learning approaches that align with the bio-inspired physical design, with a case study on a humanoid musculoskeletal head and neck robot.

For any information: Michel Verleysen -