LION11 Scope
The 11th Learning and Intelligent OptimizatioN Conference
Conference pictures
Practitioners using heuristic algorithms for hard optimization problems are confronted with the burden of selecting the most appropriate method, in many cases through expensive algorithm configuration and parameter tuning . Scientists seek theoretical insights and demand a sound experimental methodology for valuating algorithms and assessing strengths and weaknesses. This effort requires a clear separation between the algorithm and the experimenter, who, in too many cases, is "in the loop" as a motivated intelligent learning component. LION deals with designing and engineering ways of "learning" about the performance of different techniques, and ways of using past experience about the algorithm behavior to improve performance in the future. Intelligent learning schemes for mining the knowledge obtained online or offline can improve the algorithm design process and simplify the applications of high-performance optimization methods. Combinations of different algorithms can further improve the robustness and performance of the individual components.
This meeting, which continues the successful series
(1.Andalo,
2.Trento,
3.Trento,
4.Venice,
5.Rome,
6.Paris,
7.Catania,
8.Gainesville,
9.Lille,
10.Ischia), explores the intersections and uncharted territories
between machine learning, artificial intelligence, mathematical programming and algorithms for hard optimization problems.
Russia has a long tradition in optimization theory, computational mathematics and "intelligent learning techniques"
(in particular cybernetics and statistics). The location of LION11 in Nizhny is an occasion
to meet researchers and consolidate research and human links.
In internet time the value of a conference must go well beyond presenting scientific papers.
In particular, we are discussing and planning:
- Contests in continuous global optimization (a future edition of GENOPT ) and in selected discrete optimization problems.
- Sessions dedicated to commercial software and open source initiatives..
- Parallel computing, supercomputing, and cloud for "big optimization".
- Elevator pitches about new and crazy ideas (also by PhD students) to get rapid feedback by senior people.
- Tutorials about methods and software.
The conference is supported by the Russian Science Foundation (project No 15-11-30022).
Important dates
- Sep 10, 2016 call for proposals of special sessions, contests, industrial sessions (deadline Oct 10,2016)
Dec 18Dec 25, 2016 paper submission deadline.- March 10, 2017 paper acceptance notification.
- June 19-20-21, 2017, conference at Nizhny Novgorod, Russia
- Collocated with the 7th International Conference on Network Analysis : June 22-23-24, 2017
Download one-page flyer of LION11 in PDF. one-page flyer of LION11 in Word
Plenary Speakers
- Renato De Leone, Camerino, Italy
Title: The use of grossone in optimization: a survey and some recent results
Abstract: In a series of papers [1, 2], Yaroslav Sergeyev recently proposed a novel approach to infinite and infinitesimal numbers whose peculiar characteristic is the attention to numerical aspects and to applications. The numeral system is computationally and practically oriented and it is based on the use of a new infinite number called grossone and introduced as the number of elements of the set, N, of natural numbers. This approach gives the opportunity to execute numerical computations with infinite and infinitesimal numbers in a unique framework with finite quantities using a new positional numeral system with base equal to grossone. In this talk we will discuss some applications of grossone in Mathematical Programming. We will show that the new methodology with infinite and infinitesimal quantities could be used in innovative ways in penalty methods in constrained optimization, conjugate gradient methods, lexicographic multi-objective linear programming, and convex non-smooth optimization.
Keywords: Mathematical Programming; Nonlinear programming; Grossone
Related references: 1. Y.D. Sergeyev. A new applied approach for executing computations with infinite and infinitesimal quantities. Informatica, vol. 19(4), pp. 567-596 (2008)
2. Y.D. Sergeyev. Numerical computations and mathematical modelling with in finite and infinitesimal numbers. Journal of Applied Mathematics and Computing. vol. 29, pp. 177-195 (2009). - Panos Pardalos, Gainesville, USA
Title: Quantification of network dissimilarities and its practical implications
Abstract: In this lecture, we analyze a novel measure that quantifies network dissimilarities by comparing its performance with other well-known tools. The efficacy of this measure, based on Information Theory, depends on the use of rich information extracted from the graphs. We show here that the measure has promising implications in several research areas that include, bioinformatics, climate dynamics, percolation in networks, network robustness and model selection. We perform extensive computational experiments on real and artificial networks. Future research directions, which include applications to multiplex settings, will also be discussed.
This is joint work with T. Schieber, M.G. Ravetti, and L. Carpi.
Related references:
1. T.A. Schieber, L. Carpi, A. Díaz-Guilera, P.M. Pardalos, C. Masoller, M.G. Ravetti. Quantification of network structural dissimilarities . Nature Communications (2017) Jan 9;8:13928.
2. T.A. Schieber, L. Carpi, A.C. Frery, O.A. Rosso, P.M. Pardalos, and M.G. Ravetti. Information Theory Perspective on Network Robustness. Physics Letters A (2016) vol. 380(3), pp. 359-364.
3. T.A. Schieber, M.G. Ravetti, P.M. Pardalos. A Review on Network Robustness from an Information Theory Perspective . DOOR 2016, Lecture Notes in Computer Science (2016) vol. 9869, pp. 50-60. - Julius Žilinskas, Vilnius, Lithuania
Title: Deterministic algorithms for black box global optimization
Abstract: In this lecture we discuss deterministic algorithms for global optimization, where many local minimizers may exist and the objective function may be computable as a black box. A general assumption about the objective function is used that it satisfies the Lipschitz condition. We overview the branch and bound scheme for global optimization and bounds based on the Lipschitz condition. Most often the Lipschitz constant is unknown, for such a case we discuss algorithms inspired by a DIRECT scheme where a set of possible values for the Lipschitz constant is considered. We discuss various shapes of subregions including hyper rectangles and simplices, their advantages and disadvantages. We also investigate and compare bisection and trisection of subregions and various sampling strategies for evaluation of the objective function in the subregions. - Nenad Mladenovic, Belgrade, Serbia
Title: Less is more approach in Heuristic Optimization
Abstract: Heuristic approach named ’Less is more approach’ (LIMA) has recently been proposed in Mladenovic et al (2016). Its main idea is to find the minimum number of search ingredients in solving some particular optimization problem that makes some heuristic more efficient than the currently best in the literature. More precisely, the goal is to make heuristic as simple as possible, but in the same time more effective and efficient than the current state-of-the-art heuristic.
Tutorials
- Adil Erzin, Novosibirsk, Russia
Title: Some optimization problems in the wireless sensor networks
Abstract: Wireless sensor networks (WSN) consist of the electronic devices that collect information within a certain coverage area, partially process it and send data to the base station (BS). In the WSN, a critical resource is the energy of sensors, which is spent mainly on monitoring and transmitting data. In this connection, the following three main problems of computational geometry and combinatorial optimization arise.
1. Since the sensing energy consumption of the sensor is proportional to the area covered by it, the problem of energy-efficient monitoring reduces to the classical problem of constructing the least dense cover.
2. If the first problem is solved, then the sensors are placed and their parameters are specified, and for the data transmission it is necessary to build an energy-efficient communication network by determining the sensor’s transmitter powers in such a way as to obtain a connected network for the maintenance of which minimum energy is spent. This problem is known as the Min-Power Symmetric Connectivity Problem.
3. In the WSN, for transmission of data collected by sensors, as a rule, one radio frequency is used and if more than one transmitter is operating in the sensor receiving zone, then (because of the interference of radio waves) a conflict arises. When aggregating data in the BS, in order to save energy, each sensor transmits only once during the aggregation session. This means that in the communication graph, found as a result of solving problem 2, we need to find an aggregation tree (AT) rooted in the BS. Moreover, the sensor cannot simultaneously receive and transmit and cannot receive data packets from more than one sender. It is required to find an AT and a min-length conflict-free aggregation schedule.
In the tutorial the various statements of the problems 1–3, as well as algorithms for their solution, will be presented. - Mario Guarracino, Naples, Italy
Title: Laplacian-based semi supervised learning
Abstract: Supervised and unsupervised learning models are powerful tools to interpret data, with the main difference being that the former relies on the prior knowledge of class assignments (i.e. the labels) for a set of initial points, whereas in the latter a model is built without any prior information on the underlying data structure. Supervised algorithms are mostly used to produce predictive models, whereas unsupervised are more effective in characterizing data and uncovering hidden structures. Nevertheless, there is a third possibility, in which labeled and unlabeled data are used together to build models that are both characterizing and predictive. This methodology is known as semi-supervised learning. In this tutorial we will describe some novel semi-supervised techniques, providing some real world examples. - Yaroslav Sergeyev, University of Calabria (Italy), Lobachevsky University, Nizhny (Russia)
Title: Numerical computations with infinities and infinitesimals
Abstract: In this tutorial, a recent computational methodology is described. It has been introduced with the intention to allow one to work with infinities and infinitesimals numerically in a unique computational framework. It is based on the principle ‘The part is less than the whole’ applied to all quantities (finite, infinite, and infinitesimal) and to all sets and processes (finite and infinite). The methodology uses as a computational device the Infinity Computer (patented in USA and EU) working numerically with infinite and infinitesimal numbers that can be written in a positional system with an infinite radix. On a number of examples (numerical differentiation and optimization, divergent series, ordinary differential equations, fractals, set theory, etc.) it is shown that the new approach can be useful from both theoretical and computational points of view. The accuracy of the obtained results is continuously compared with results obtained by traditional tools used to work with mathematical objects involving infinity.
Paper Submission deadline December, 25
Paper Format
Please prepare your paper in English using the Lecture Notes in Computer Science (LNCS) template, which is available here . Papers must be submitted in PDF.Types of Submissions
When submitting a paper to LION11, authors are required to select one of the following three types of papers:- Long paper: original novel and unpublished work (max. 15 pages in LNCS format);
- Short paper: an extended abstract of novel work (max. 6 pages in LNCS format);
- Work for oral presentation only (no page restriction; any format). For example, work already published elsewhere, which is relevant and which may solicit fruitful discussion at the conference.
Submission System
All papers must be submitted using EasyChair.
Registration
Participants of the conference have to register on the website till April 22.
The conference is organized in the Oka hotel . We recommend to reserve an accommodation in this hotel.
You could select a room in the registration form (please, specify arrival/departure dates). The Organizing Committee will reserve accommodation for you for these dates. Payment can be done at check-in. After April, 22, the Organizing Committee is not responsible for the accommodation for the conference participants.
You could reserve accommodation independently using website of the hotel or booking.com service ( Oka 3*, Oka 4* )
Please, do not apply for a tourist visa! The visa application must be submitted in strict accordance with the invitation from Nizhny Novgorod State University.
Regular registration fee (till
- 300 Euro
- 220 Euro for master and postgraduate students
- 400 Euro
- 250 Euro for master and postgraduate students
