- Lisa_seminaires - mailman.iro.umontreal.ca

Fwd: next math-neuro: 27th Jan (Tuesday)
by Dumitru Erhan 26 Jan '09

26 Jan '09

Begin forwarded message: > From: Sergiy Yakovenko <sergiy.yakovenko(a)gmail.com> > Date: January 22, 2009 9:02:20 PM EST (CA) > To: Undisclosed-recipients: <>; > Cc: Timothy Lillicrap <tim(a)biomed.queensu.ca> > Subject: next math-neuro: 27th Jan (Tuesday) > > Dear Math-Neuro group, > > Our next Math-Neuro meeting will be next week on 27th of January (Tuesday), 4-6 PM, room 4106 of Pav. P.-G. Desmarais. Timothy Lillicrap of Queen's University will present an exciting talk entitled "How do the properties of the musculoskeletal system shape the activity of motor cortex neurons?" > > Regards, > Sergiy & Paul > > Sergiy Yakovenko > Département de physiologie > Université de Montréal > Tel: (514) 343-6111 ext 3333 > > DIRECTIONS: > http://maps.google.com/maps/ms?client=safari&oe=UTF-8&ie=UTF8&hl=en&t=k&lci… > > > >

1 0

UdeM-McGill-MITACS machine learning seminar Fri January 30th, 15:00,
by Dumitru Erhan 26 Jan '09

26 Jan '09

This week's seminar (see http://www.iro.umontreal.ca/article.php3?id_article=107&lang=en) New architectures and algorithms for solving deep-memory POMDPs by Daan Wierstra, IDSIA, Switzerland Location: Pavillon André-Aisenstadt (UdeM), room 3195 Time: January 30th 2009, 15h00 Reinforcement learning in partially observable environments is a difficult problem, and appears to be intractable for classical reinforcement learning methods. We present two new approaches for dealing with partial observability. One pertains to searching the space of memory cell architectures in order to better capture long-term time dependencies. The second approach involves the use of natural gradients in stochastic search as an alternative to conventional evolutionary methods.

1 0

[Fwd: COLLOQUE CRM-ISM-GERAD DE STATISTIQUE (05/12/2008, Peter Hoff)]
by Yoshua Bengio 25 Nov '08

25 Nov '08

Ce séminaire devrait être intéressant pour les chercheurs en machine learning... -------- Original Message -------- Subject: COLLOQUE CRM-ISM-GERAD DE STATISTIQUE (05/12/2008, Peter Hoff) Date: Tue, 25 Nov 2008 12:53:02 -0500 From: CRM(a)CRM.UMontreal.CA To: activites(a)CRM.UMontreal.CA ****************************************************************** COLLOQUE CRM-ISM-GERAD DE STATISTIQUE ****************************************************************** CONFERENCIER(S) / SPEAKER(S) : Peter Hoff (University of Washington) TITRE / TITLE : Hierarchical eigenmodels for pooled covariance estimation LIEU / PLACE : CRM, UdeM, Pav. André-Aisenstadt, 2920, ch. de la Tour, salle 5340 DATE : Le vendredi 5 décembre 2008 / Friday, December 5, 2008 HEURE / TIME : 15 h 30 / 3:30 p.m. RESUME / ABSTRACT : While a set of covariance matrices corresponding to different populations are unlikely to be exactly equal, they can still exhibit a high degree of similarity. For example, some pairs of variables may be positively correlated across most groups, while other pairs may be consistently negative. In such cases the similarities across covariance matrices can be described by similarities in their principal axes, the axes defined by the eigenvectors of the covariance matrices. Estimating the degree of across-population eigenvector heterogeneity can be helpful for a variety of estimation tasks. Similar eigenvector matrices can be pooled to form a central set of principal axes, and covariance estimation for populations having small sample sizes can be stabilized by shrinking estimates of their population-specific principal axes towards the across-population center. To this end, in this talk we'll discuss a hierarchical model and estimation procedure for pooling principal axes across several populations. The model for the across-group heterogeneity is based on a matrix valued antipodally symmetric Bingham distribution that can flexibly describe notions of center and spread for a population of orthonormal matrices. Une réception suivra (salle 4361) / Cocktail will follow (Room 4361). ****************************************************************** Responsable : Alejandro Murua (murua(a)dms.umontreal.ca, (514-343-6987) http://www.crm.umontreal.ca/cgi/Stats.csh ****************************************************************** -- Yoshua Bengio

1 0

Reminder - Tomorrow 15:00 - Talk by Mathieu Sinn
by Pascal Lamblin 20 Nov '08

20 Nov '08

Hi, Don't forget tomorrow's MITACS-Lisa-McGill seminar! Title: Analysis of EEG data by means of ordinal pattern distributions Speaker: Mathieu Sinn Date: Friday, November 21st, 15:00 Place: Room 3195 (pav. André-Aisenstadt, UdeM) Abstract: Ordinal time series analysis is a new fast and robust approach to the investigation of long and complex time series, such as electroencephalographic (EEG) data. The basic idea is to consider locally the order relations between the values instead of the values themselves. An ordinal pattern represents the order relations between a finite number of equidistant values of a time series. Assuming that these values are pairwise different, their unique rank order can be simply described by a permutation. By looking at the distributions of ordinal patterns (or measures derived from the distributions) in different parts of a time series, changes in the dynamic of the underlying system can be detected, such as increasing or decreasing complexity. One derived measure is the permutation entropy, which is the Shannon entropy of ordinal pattern distributions. It has been shown that in certain cases of time series obtained by dynamical systems, the permutation entropy is related to the Kolmogorov-Sinai entropy which is a well-known measure for the complexity of time series and the underlying systems, respectively. A main application of the permutation entropy so far has been the detection of epileptic activity in EEG data, as it is widely accepted that the complexity of the brain dynamics changes (namely, decreases) during epileptic seizures. However, since clearly different ordinal pattern distributions may yield very similar permutation entropy values, we propose to take the whole distributions into account. First attempts show that a cluster analysis of ordinal pattern distributions may be a good approach to an automatic separation and classification of EEG data. Beneath the detection and localization of epileptic activity, another interesting application of this method is the automatic classification of sleep stages according to the criterions of Rechtschaffen and Kales. This is joint work with Karsten Keller. -- Pascal

1 0

Reminder - Tomorrow 15:00 - Talk by Rion Snow
by Pascal Lamblin 16 Nov '08

16 Nov '08

Just to remind you, because Monday is tomorrow: Title: Cheap and Fast But is it Good? Evaluating Nonexpert Annotations for Machine Learning Tasks Speaker: Rion Snow Date: Monday, November 17th, 15:00 Place: Room 3195 (pav. André-Aisenstadt, UdeM) Abstract: Human annotation is crucial for many machine learning tasks but can be expensive and time-consuming. We explore the use of Amazon's Mechanical Turk web service, a significantly cheaper and faster method for collecting annotations from a broad base of paid non-expert contributors over the Web. We investigate five task in the field of natural language processing: affect recognition, word similarity, recognizing textual entailment, event temporal ordering, and word sense disambiguation. For all five, we show high agreement between Mechanical Turk non-expert annotations and existing gold standard labels provided by expert labelers. For the task of affect recognition, we also show that using non-expert labels for training machine learning algorithms can be as effective as using gold standard annotations from experts. We propose a technique for bias correction that significantly improves annotation quality on two tasks. We conclude that many large labeling tasks can be effectively designed and carried out in this method at a fraction of the usual expense. A summary of this work may be found online at: http://blog.doloreslabs.com/2008/09/amt-fast-cheap-good-machine-learning/. Bio: Rion Snow is a PhD Candidate in Computer Science at Stanford University, advised by Professors Andrew Ng and Dan Jurafsky. Rion works in the intersection of machine learning and natural language processing, with a focus in computational semantics. He leads the Stanford Wordnet Project, which aims at learning large-scale semantic networks automatically from natural text. His work on automatically inferring semantic taxonomies received the Best Paper Award at the 2006 conference for the Association of Computational Linguistics. His publications and contact information may be found at his home page: http://ai.stanford.edu/~rion/. -- Pascal

1 0

Next week - 2 UdeM-McGill-MITACS seminars!
by Pascal Lamblin 12 Nov '08

12 Nov '08

Hi everyone, Next week, we're lucky to have two invited speakers giving a talk at UdeM, the first on Monday, the second on Friday, both at 15:00. Descriptions of both talks follow. ---- Title: Cheap and Fast But is it Good? Evaluating Nonexpert Annotations for Machine Learning Tasks Speaker: Rion Snow Date: Monday, November 17th, 15:00 Place: Room 3195 (pav. André-Aisenstadt, UdeM) Abstract: Human annotation is crucial for many machine learning tasks but can be expensive and time-consuming. We explore the use of Amazon's Mechanical Turk web service, a significantly cheaper and faster method for collecting annotations from a broad base of paid non-expert contributors over the Web. We investigate five task in the field of natural language processing: affect recognition, word similarity, recognizing textual entailment, event temporal ordering, and word sense disambiguation. For all five, we show high agreement between Mechanical Turk non-expert annotations and existing gold standard labels provided by expert labelers. For the task of affect recognition, we also show that using non-expert labels for training machine learning algorithms can be as effective as using gold standard annotations from experts. We propose a technique for bias correction that significantly improves annotation quality on two tasks. We conclude that many large labeling tasks can be effectively designed and carried out in this method at a fraction of the usual expense. A summary of this work may be found online at: http://blog.doloreslabs.com/2008/09/amt-fast-cheap-good-machine-learning/. Bio: Rion Snow is a PhD Candidate in Computer Science at Stanford University, advised by Professors Andrew Ng and Dan Jurafsky. Rion works in the intersection of machine learning and natural language processing, with a focus in computational semantics. He leads the Stanford Wordnet Project, which aims at learning large-scale semantic networks automatically from natural text. His work on automatically inferring semantic taxonomies received the Best Paper Award at the 2006 conference for the Association of Computational Linguistics. His publications and contact information may be found at his home page: http://ai.stanford.edu/~rion/. ---- Title: Analysis of EEG data by means of ordinal pattern distributions Speaker: Mathieu Sinn Date: Friday, November 21st, 15:00 Place: Room 3195 (pav. André-Aisenstadt, UdeM) Abstract: Ordinal time series analysis is a new fast and robust approach to the investigation of long and complex time series, such as electroencephalographic (EEG) data. The basic idea is to consider locally the order relations between the values instead of the values themselves. An ordinal pattern represents the order relations between a finite number of equidistant values of a time series. Assuming that these values are pairwise different, their unique rank order can be simply described by a permutation. By looking at the distributions of ordinal patterns (or measures derived from the distributions) in different parts of a time series, changes in the dynamic of the underlying system can be detected, such as increasing or decreasing complexity. One derived measure is the permutation entropy, which is the Shannon entropy of ordinal pattern distributions. It has been shown that in certain cases of time series obtained by dynamical systems, the permutation entropy is related to the Kolmogorov-Sinai entropy which is a well-known measure for the complexity of time series and the underlying systems, respectively. A main application of the permutation entropy so far has been the detection of epileptic activity in EEG data, as it is widely accepted that the complexity of the brain dynamics changes (namely, decreases) during epileptic seizures. However, since clearly different ordinal pattern distributions may yield very similar permutation entropy values, we propose to take the whole distributions into account. First attempts show that a cluster analysis of ordinal pattern distributions may be a good approach to an automatic separation and classification of EEG data. Beneath the detection and localization of epileptic activity, another interesting application of this method is the automatic classification of sleep stages according to the criterions of Rechtschaffen and Kales. This is joint work with Karsten Keller. ---- Hoping to see you there, -- Pascal

1 0

Nov 11th, 10:30 - UdeM-McGill-MITACS seminar by Tijmen Tieleman
by Pascal Lamblin 07 Nov '08

07 Nov '08

Hello everyone, and sorry for the late notice, Next MITACS seminar of will be on Tuesday, November 11th, 10:30 at UdeM (room AA-3195). WARNING: Notice the unusual time and day! Title: Using fast weights to improve Persistent Contrastive Divergence Speaker: Tijmen Tieleman Date: Tuesday, November 11th, 10:30 (a.m.) Place: Room 3195 (pav. André-Aisenstadt, UdeM) Abstract: The most commonly used learning algorithm for restricted Boltzmann machines is contrastive divergence which starts a Markov chain at a data point and runs the chain for only a few iterations to get a cheap, low variance estimate of the sufficient statistics under the model. Tieleman (2008) showed that better learning can be achieved by estimating the model's statistics using a small set of persistent "fantasy particles" that are not reinitialized to data points after each weight update. With sufficiently small weight updates, the fantasy particles represent the equilibrium distribution accurately but to explain why the method works with much larger weight updates it is necessary to consider the interaction between the weight updates and the Markov chain. We show that the weight updates force the Markov chain to mix fast, and using this insight we develop an even faster mixing chain that uses an auxiliary set of "fast weights" to implement a temporary overlay on the energy landscape. The fast weights learn rapidly but also decay rapidly and do not contribute to the normal energy landscape that defines the model. This is joint work with Geoffrey Hinton. I Hope to see you there, -- Pascal

1 0

FW: CERNEC Conférence - Vendredi 17 Oct. (Dr. Marc Schoenwiesner)
by Douglas Eck 14 Oct '08

14 Oct '08

Here's a really interesting talk. Unfortunately it conflicts with Geoff Hinton's talk. . . . but I wanted to announce it anyway. -Doug ---------- Forwarded message ---------- From: Schoenwiesner Marc <marc.schoenwiesner(a)umontreal.ca> Date: 2008/10/14 Subject: [BRAMS-local] FW: CERNEC Conférence - Vendredi 17 Oct. (Dr. Marc Schoenwiesner) To: brams-local(a)iro.umontreal.ca Bonjour, Vous êtes cordialement invités à la prochaine conférence du Centre de Recherche en Neuropsychologie et Cognition (CERNEC). Hi, You are cordially invited to the next Centre de Recherche en Neuropsychologie et Cognition (CERNEC) conference. Conférencier / Speaker : Dr. Marc Schoenwiesner 'The human hearing system - what we know, what we want to know, and how we can find out' Date : Vendredi 17 octobre 2008, 15h00 Lieu : Local D-427 du pavillon Marie-Victorin de l'Université de Montréal, 90 avenue Vincent d'Indy (à la sortie du métro Édouard-Montpetit)

1 0

Oct 16th, 15:00 - UdeM-McGill-MITACS seminar by Geoffrey Hinton
by Pascal Lamblin 10 Oct '08

10 Oct '08

The second MITACS seminar of the fall will be on Thursday, October 16th, 15:00 at UdeM. Title: Visualizing high-dimensional data using t-SNE Speaker: Geoffrey Hinton Date: Thursday Oct 16th, 2008, 15:00 Place: Room 3195 (pav. André-Aisenstadt) Abstract: Over the last decade, many new methods have been developed for visualizing high-dimensional data by giving each data-point a location in a two-dimensional map. The goal is to represent the separations of pairs of data-points by the separations of their corresponding map-points, with an emphasis on representing the small separations accurately. I will describe a new method, called t-SNE, that is based on two ideas. The first idea is to convert the set of pairwise distances between data-points into a set of probabilities of selecting pairs of data-points. The selection probability of a pair of points is proportional to a Gaussian function of their separation. If the distances between map-points are converted into pairwise probabilities in the same way, any given arrangement of map-points can be evaluated by measuring the divergence between the probability distributions obtained from the data-points and the map-points. A good arrangement of map-points is then found by performing gradient descent in this divergence. Unfortunately, if the probabilities of pairs of map-points are computed using a Gaussian function of their separation, the difference between the distributions of pairwise distances in high-dimensional and low-dimensional spaces causes the map-points to be crowded together in the center of the map. This problem can be largely overcome by using a heavy-tailed t-distribution when computing the selection probabilities of pairs of map-points. This leads to maps that look much better than those produced by other recent methods. In particular, t-SNE is very good at preserving clusters in the data at many different scales simultaneously. The talk describes joint work with Laurens van der Maaten that will appear in the Journal of Machine Learning Research. -- Pascal

1 1

Oct 17th, 14:30 talk by Hinton at McGill - The next generation of neural networks
by Pascal Lamblin 07 Oct '08

07 Oct '08

Geoff Hinton is giving a talk next week (friday, oct. 17th) at McGill, organized by CRM Applied Math Lab and McGill CSE, <http://www.dms.umontreal.ca/~mathapp/> Location: McGill, Burnside Hall 1205 Time: 14:30 Speaker: Geoffrey Hinton <http://www.cs.toronto.edu/~hinton/> Title: THE NEXT GENERATION OF NEURAL NETWORKS Abstract: In the 1980's, new learning algorithms for neural networks promised to solve difficult classification tasks, like speech or object recognition, by learning many layers of non-linear features. The results were disappointing for two reasons: There was never enough labeled data to learn millions of complicated features and the learning was much too slow in deep neural networks with many layers of features. These problems can now be overcome by learning one layer of features at a time and by changing the goal of learning. Instead of trying to predict the labels, the learning algorithm tries to create a generative model that produces data which looks just like the unlabeled training data. After learning many layers of features in this way, a relatively small amount of labeled data can then be used to fine-tune the features to give better discrimination. These new neural networks outperform other machine learning methods when labeled data is scarce but unlabeled data is plentiful. I will describe an application to recognizing 3-D shapes by Vinod Nair and an application to generating human motion by Graham Taylor. -- Pascal

1 0