Computational Information Geometry Wonderland

Olivier attended the ICPR conference last August in Turkey. He presented the work on Burbea-Rao centroids. Here is an overview of the conference. Frank.

ICPR 2010 -- International Conference on Pattern Recognition

The 20th edition of ICPR held in Istanbul, Turkey from 23rd to 26th August 2010. The size of the event was quite impressive: around 1200 participants presenting more than one thousand papers, 385 oral presentations and 777 posters, but still being selective since more than 2100 papers were submitted. The drawback may have been the number of parallel sessions, but given the number of people and the diversity of fields it was unlikely to see nearly empty rooms or to miss interesting papers in your main research topic.

As usual, we saw a lot of famous invited researchers during plenary sessions. Christopher M. Bishop (Microsoft Research Cambridge, UK) in a talk entitled /Embracing Uncertainty: The New Machine Intelligence/ made a kind of history of machine learning from expert driven approaches to black-box models and presented graphical models as a way to combine these two points of view. Shree K. Nayar (Columbia University, USA) in /Computational Cameras: Redefining the Image/ presented new way of defining photography by putting intelligence directly in the camera and thus allowing to improve quality of images or opening the path to completely new applications and ways to see our world. Prabhakar Raghavan (Yahoo! Research, USA) gave a talk with the title /The Quantitative analysis of user behavior online data models and algorithms/ and presented some ideas about interaction between computer science and social sciences, in order to study the way users react to content presented to them. The invited speakers plenary sessions were completed by award winner sessions: Horst Bunke (University of Bern, Switzerland) for /Towards the Unification of Structural and Statistical Pattern Recognition/ and Antonio Torralba for /Scene and Object Recognition in Context/.

Due to our research interests, I mainly followed the classification, clustering and retrieval tracks. Given the number of papers and thus the number of interesting talks or posters, I will need to present a very short selection of these interesting papers, which will be necessarily incomplete and unfair. In /Saliency based on Multi-Scale Ratio of Dissimilarity/ Hang et al. presented a new saliency map building method which gives more reasonable results and preserves the resolution of the input image. /Effective Dimensionality Reduction based on Support Vector Machine/ by Moon et al. describes a dimension reduction algorithm which exploits the fact that a SVM minimizes the structural risk. Rudinac and Jonker presented the paper /Saliency Detection and Object Localization in Indoor Environments/, introducing a cheap way of computing saliency maps which is affordable for embedded systems. /Maximum Likelihood Estimation of Gaussian Mixture Models using Particle Swarm Optimization/ by Ari and Aksoy showed an effective use of a population-based algorithm to compute a mixture. In /Improving the Efficiency of Content-Based Multimedia Exploration/ Beecks et al. present a new way of presenting search results allowing users to perform queries intuitively. /High-Level Feature Extraction using SIFT GMMs and Audio Models/ by Inoue et al. used mixture models of SIFT descriptors and hidden markov models with good performances on the TRECVID09 dataset. /Characterising Facial Gender Difference using Fisher-Rao Metric/ by Ceolin presented an effective example of the use of the Fisher-Rao distance.

The Istanbul conference center and the organizing team were up to the event. The scientific part was completed by two social events: a welcome reception in the garden of the Bogazici University and a banquet in the prestigious Ciragan Palace Kempinski, an ancient Ottoman imperial palace transformed in a luxurious hotel. Last but not least, the stay in Istanbul was very pleasant at this end of the month of August.

By Olivier.

Yet another report by Olivier who attended RFIA 2010

RFIA 2010 -- Reconnaissance de formes et intelligence artificielle

RFIA (which stands for Reconnaissance de formes et intelligence artificielle) is a French-speaking conference about pattern recognition and artificial intelligence. The 17th edition held in Caen from January 20 to January 22 2010.

There was four parts in the workshop: an invited speaker session, a short talk session, a poster session and a demo session.

The short talk session was divided into two tracks: pattern recognition and artificial intelligence. I mainly followed the Pattern Recognition track. Unfortunately I was only able to attend the first two days of the conference, so I will present only talks from these days. Sang Ly presented a method to estimate moves of a stereoscopic vision system. Hervé Jégou presented a new compact representation of bag-of-words for image retrieval. Thierry Germa presented some results to follow people with a mobile platform, using information from image and RFID tags. Pierre Lébraly explained an extrinsic calibrating method for multi-camera system, using a planar mirror.

The three invited talks I attended were given by Olivier Teytaud (INRIA Saclay, France) speaking about artificial intelligence methods for Go game, Schlomo Zilberstein (University of Massachusetts, USA) presenting challenges and directions for decentralized decision making and by Nicu Sebe (University of Amsterdam, Nederlands) detailing some perspectives for human centered computing.

Frank.