- Lisa_teatalk - mailman.iro.umontreal.ca

No tea-talk today
by Guillaume Desjardins 21 Mar '13

21 Mar '13

There is unfortunately no tea-talk scheduled for today. On a related note, we cannot have tea-talks without volunteers. So if you haven't yet presented an ICLR paper, please do so. We have open spots for the next 2-3 weeks. -- Guillaume

1 0

Tea-talk: today, Thursday March 14th @ 2pm, lisa lab.
by Guillaume Desjardins 14 Mar '13

14 Mar '13

Sorry for the late notice. Please join us for today's tea-talk at 2pm. Speakers and titles below. On Wed, Mar 6, 2013 at 8:52 AM, Guillaume Desjardins < guillaume.desjardins(a)gmail.com> wrote: > This week's tea-talk will be postponed to next week (Thursday, March > 14th), as several students are away for reading week and one of the > speakers is feeling a bit under the weather. > > Next week's speakers will be: > * Ian Goodfellow, who will discuss a recent ICML submission on "Maxout > Networks" <http://arxiv.org/abs/1302.4389> > * Yann Dauphin, who will discuss his ICLR submission, "Big Neural > Networks Waste Capacity<http://openreview.net/document/5563cdeb-e853-47ac-83b2-1b1b20cc1535#5563cde…> > " > > -- > Guillaume >

1 0

No tea-talk this week
by Guillaume Desjardins 06 Mar '13

06 Mar '13

This week's tea-talk will be postponed to next week (Thursday, March 14th), as several students are away for reading week and one of the speakers is feeling a bit under the weather. Next week's speakers will be: * Ian Goodfellow, who will discuss a recent ICML submission on "Maxout Networks" <http://arxiv.org/abs/1302.4389> * Yann Dauphin, who will discuss his ICLR submission, "Big Neural Networks Waste Capacity<http://openreview.net/document/5563cdeb-e853-47ac-83b2-1b1b20cc1535#5563cde…> " -- Guillaume

1 0

Tea-talk: Thursday, Feb 28th @ 2pm (lisa lab)
by Guillaume Desjardins 27 Fév '13

27 Fév '13

Razvan and I will be discussing our recent ICLR submissions tomorrow afternoon at 2:00pm. We have discussed very similar topics in a past tea-talks, but hopefully things which be much clearer (and convincing) now that the dust has settled a little bit. Next week, we will have Ian and Caglar discussing their ICML (maxout) and ICLR submissions respectively. More volunteers will be needed in 2 weeks for discussing other ICLR submissions. See you there ! Title: Natural Gradient Revisited<http://openreview.net/document/d54212e3-fa9a-40c4-991d-320bda524718#d54212e…> Razvan Pascanu, Yoshua Bengio The aim of this paper is two-folded. First we intend to show that Hessian-Free optimization (Martens, 2010) and Krylov Subspace Descent (Vinyals and Povey, 2012) can be described as implementations of Natural Gradient Descent due to their use of the extended Gauss-Newton approximation of the Hessian. Secondly we re-derive Natural Gradient from basic principles, contrasting the difference between the two version of the algorithm that are in the literature. Title: Metric-Free Natural Gradient for Joint-Training of Boltzmann Machines<http://openreview.net/document/88a8fd8a-5af9-4189-9059-9749a5da5bdc#88a8fd8…> Guillaume Desjardins, Razvan Pascanu, Aaron Courville, Yoshua Bengio This paper introduces the Metric-Free Natural Gradient (MFNG) algorithm for training Boltzmann Machines. Similar in spirit to the Hessian-Free method of Martens [8], our algorithm belongs to the family of truncated Newton methods and exploits an efficient matrix-vector product to avoid explicitely storing the natural gradient metric $L$. This metric is shown to be the expected second derivative of the log-partition function (under the model distribution), or equivalently, the variance of the vector of partial derivatives of the energy function. We evaluate our method on the task of joint-training a 3-layer Deep Boltzmann Machine and show that MFNG does indeed have faster per-epoch convergence compared to Stochastic Maximum Likelihood with centering, though wall-clock performance is currently not competitive.

1 0

Deep Mind Tea-talks: Wed., Feb 20th @ 1h30pm (AA3256) + Thurs. Feb 21st @ 2:00pm (AA3256)
by Guillaume Desjardins 21 Fév '13

21 Fév '13

Hi everyone, We have two upcoming tea-talks by Daan Wierstra and Karol Gregor of Deep Mind. Dan Wierstra will start tomorrow (Wed. Feb 20th, @1:30pm) by providing an overview of Deep Mind and their ongoing work in deep learning. On Thursday at 2:00pm, Karol Gregor will present two of his recent papers (title and abstract below). See you there ! Title: A lattice filter model of the visual pathway Authors: Karol Gregor, Dmitri B Chklovskii Abstract: Early stages of visual processing are thought to decorrelate, or whiten, the incoming temporally varying signals. Because the typical correlation time of natural stimuli, as well as the extent of temporal receptive fields of lateral geniculate nucleus (LGN) neurons, is much greater than neuronal time constants, such decorrelation must be done in stages combining contributions of multiple neurons. We propose to model temporal decorrelation in the visual pathway with the lattice filter, a signal processing device for stage-wise decorrelation of temporal signals. The stage-wise architecture of the lattice filter maps naturally onto the visual pathway (photoreceptors -> bipolar cells -> retinal ganglion cells -> LGN) and its filter weights can be learned using Hebbian rules in a stage-wise sequential manner. Moreover, predictions of neural activity from the lattice filter model are consistent with physiological measurements in LGN neurons and fruit fly second-order visual neurons. Therefore, the lattice filter model is a useful abstraction that may help unravel visual system function. Title: Fast Approximations to Structured Sparse Coding and Applications to Object Classification Authors: Arthur Szlam , Karol Gregor, and Yann LeCun Abstract: We describe a method for fast approximation of sparse coding. A given input vector is passed through a binary tree. Each leaf of the tree contains a subset of dictionary elements. The coefficients corresponding to these dictionary elements are allowed to be nonzero and their values are calculated quickly by multiplication with a precomputed pseudoinverse. The tree parameters, the dictionary, and the subsets of the dictionary corresponding to each leaf are learned. In the process of describing this algorithm, we discuss the more general problem of learning the groups in group structured sparse modeling. We show that our method creates good sparse representations by using it in the object recognition framework of [1,2]. Implementing our own fast version of the SIFT descriptor the whole system runs at 20 frames per second on 321 × 481 sized images on a laptop with a quad-core cpu, while sacrificing very little accuracy on the Caltech 101, Caltech 256, and 15 scenes benchmarks.

1 2

Best of NIPS tea-talks (round 2): Feb. 7th, 3:00PM (lisa labo)
by Guillaume Desjardins 07 Fév '13

07 Fév '13

We will continue our best of NIPS series tomorrow, with Pascal and Ian. They will be presenting the following papers: Pascal: "A Stochastic Gradient Method with an Exponential Convergence Rate for Finite Training Sets Nicolas Le Roux, Mark Schmidt, Francis Bach. Ian: "Practical Bayesian Optimization of Machine Learning Algorithms" Jasper Snoek, Hugo Larochelle, Ryan P. Adams + my experiences trying out Jasper's spearmint-lite software We will take a break next week because of ICML, but will be back the week after for one final talk by Yann Dauphin (That's right Yann, we haven't forgotten you !), -- Guillaume On Mon, Jan 28, 2013 at 5:00 PM, Guillaume Desjardins <guillaume.desjardins(a)gmail.com> wrote: > > Tea-talks will resume this week, with our "best of NIPS" series. > Speakers this week will be: > > * Çağlar, presenting "Discriminative Learning of Sum-Product Networks" > by Robert Gens and Pedro Domingos. > * Eric, presenting "Learning Label Trees for Probabilistic Modelling > of Implicit Feedback" by Andriy Mnih, Yee Whye Teh. > > Talks will be around 30mins each (approx. 20 mins for talk and 10 mins > for questions). > > -- > Guillaume

3 3

Best of NIPS tea-talks: Thurs. Jan 31st, 3:00PM (lisa labo)
by Guillaume Desjardins 28 Jan '13

28 Jan '13

Tea-talks will resume this week, with our "best of NIPS" series. Speakers this week will be: * Çağlar, presenting "Discriminative Learning of Sum-Product Networks" by Robert Gens and Pedro Domingos. * Eric, presenting "Learning Label Trees for Probabilistic Modelling of Implicit Feedback" by Andriy Mnih, Yee Whye Teh. Talks will be around 30mins each (approx. 20 mins for talk and 10 mins for questions). -- Guillaume

1 0

Tea-talk by David Warde-Farley and Razvan Pascanu on motion-compression, Thursday, Dec 13th @ 3pm. (AA3256)
by Guillaume Desjardins 13 Déc '12

13 Déc '12

Please join us this Thursday, for a tea-talk by David and Razvan on motion compression. Title and abstract below. Razvan and I will talk about our current pipeline for human motion compression, based on interpolation, L1 sparse coding, fine tuning and a bit of elbow grease. It achieves compression ratios comparable with the state of the art and very good fidelity (any artifacts are virtually imperceptible), though not quite the rock bottom distortion rates that the current state of the art work reports. We'd like your feedback on the best way to "sell" this work for publication. (Joint work with Razvan Pascanu, Steven Pigeon and Yoshua)

1 1

Tea-talk: today, Nov 29th, 3pm (LISA lab)
by Guillaume Desjardins 29 Nov '12

29 Nov '12

Please join us for today's tea-talk. Caglar will be doing a practice talk for his upcoming workshop poster at the NIPS Deep Learning Workshop. Time-willing, I will then present the paper entitled "Cardinality Restricted Boltzmann Machines" by Kevin Swersky et al. We may also have special appearances by Ian Goodfellow and Razvan Pascanu. Abstract (Caglar Gulcehre): We explore the effect of introducing prior information into the intermediate level of deep learning algorithms for a learning task on which all the state-of-the-art machine learning algorithms tested failed to learn. We motivate our work from the hypothesis that humans learn such intermediate concepts from other individuals via a form of supervision or guidance using a curriculum. The experiments we have conducted provide positive evidence in favor of this hypothesis. In our experiments, a two-tiered MLP architecture is trained on a dataset with 64x64 binary inputs images, each image with three sprites. The ﬁnal task is to decide whether all the sprites are the same or one of them is different. Sprites are pentomino tetris shapes and they are placed in an image with different locations using scaling and rotation transformations. The ﬁrst level of the two-tiered MLP is pretrained with intermediate level targets being the presence of sprites at each location, while the second level takes the output of the ﬁrst level as input and predicts the ﬁnal task target binary event. The two-tiered MLP architecture, with a few tens of thousand examples, was able to learn the task perfectly, whereas all other algorithms (include unsupervised pre-training, but also traditional algorithms like SVMs, decision trees and boosting) all perform no better than chance. We hypothesize that the optimization difﬁculty involved when the intermediate pre-training is not performed is due to the composition of two highly non-linear tasks. Our ﬁndings are also consistent with hypotheses on cultural learning inspired by the observations of optimization problems with deep learning, presumably because of effective local minima.

1 0

LF: tea-talk volunteers for Thursday Nov 29th, 3pm
by Guillaume Desjardins 25 Nov '12

25 Nov '12

It is a tradition in the LISA lab, for students attending NIPS, to present a paper of their choosing in front of the lab. These papers should be selected based on their merits, your interests (obviously) but also those of the lab at large. This year, we have decided that a subset of these talks will given *prior* to actually attending the conference. This has many advantages: being familiar with the list of accepted papers (especially the best ones) can help guide you during the conference and knowing the material beforehand will also enrich your interactions with other researchers, etc. For this Thursday, I am thus looking for 3 volunteers to perform 20 mins tea-talks on a paper of your choosing. The list of accepted papers has been published here http://nips.cc/Conferences/2012/Program/accepted-papers.php. Papers should be available online on the authors' webpage. Everyone attending NIPS will end up presenting (either before or after the conference). If you go first, there are thus more chances of you being able to present a paper of your choosing ! You can use the following spreadsheet to coordinate who is presenting what: https://docs.google.com/spreadsheet/ccc?key=0AguEeBPRuaM2dGR1OGl5VU5Lb1RkZl… Thanks ! Guillaume

3 2

Tea-Talk by Gregoire Mesnil, Thursday, Nov 22nd @ 3pm
by Guillaume Desjardins 22 Nov '12

22 Nov '12

Slight change of plan for next week's tea-talk. Gregoire will be giving us an overview of his recent work at Microsoft Research. The talk will be held Thursday, Nov 22nd at 3pm (as usual). David (and Razvan's) tea-talk on motion compression will be moved to the week after NIPS (December 13th). ==== An abstract of what has been done this summer with Xiaodong He at Microsoft Research Redmond. There is Sequential Neural Network in this work and I would appreciate the feedback of experts in RNN to know if it's actually new or not. We plan to publish this work soon. Here it goes: Abstract: Neural Network for Spoken Language Understanding (SLU) One of the key problems in spoken language understanding is slot-filling/concept extraction. During this summer, we have worked on this problem using neural networks from three aspects. 1) at the frame level, we built a NN based slot classification model. When initialized with pre-trained word embeddings (SENNA), it gives significant improvement over logistic regression models using n-gram features. 2) at the sequence level, we developed a new (?) sequential NN (SeqNN) model, which is extended from the Recurrent NN, to take into account sequential dependencies. When evaluated on the ATIS data set, it outperforms the CRF baseline significantly. 3) we further studied the robustness of the SeqNN model under noisy/adverse condition. We will present the newly created bi-lingual (En-Zh) ATIS database, and discuss the model’s performance under a cross-lingual SLU setting with “Bing translator” noise. We will conclude the presentation with comprehensive experimental results and discussion of future works.

1 1

mandatory viewing for all
by Yoshua Bengio 16 Nov '12

16 Nov '12

A great talk by Geoff Hinton, which you should all view! http://www.youtube.com/watch?v=DleXA5ADG78 -- Yoshua

1 0

Tea-talk: today, Thurs. Nov 8th @3PM
by Guillaume Desjardins 08 Nov '12

08 Nov '12

Sorry for the late notice, but there will be a tea-talk (as usual) this afternoon at 3PM. We will continue on the topic of Deep Boltzmann Machines, but this time, exploring how the natural gradient can help train DBMs jointly. The talk will be given by Razvan and myself. -- Guillaume

1 0

Tea Talk: Thursday, November 1st @ 3PM, location AA3195
by Guillaume Desjardins 31 Oct '12

31 Oct '12

Please join us tomorrow for a tea-talk by Ian Goodfellow. When: Nov 1st, 3pm Where: AA3195 Title: Ian's Ramblings on Deep Boltzmann Machines Topics will include joint-training, including a new deterministic in-painting criteria. We will follow up over the next few weeks with the following tea-talks: * Razvan & Guillaume: joint-training of DBM through Natural Gradient. * David & Razvan, on their work on compression. * Sina Honari, who will present the Deep Lambertian Networks. * Gregoire Mesnil, who will present his work at Microsoft Research. Please start thinking about your own presentation topics, as I will need more volunteers once this queue dries out. Thanks, Guillaume

1 0

Fwd: Tea Talk: Thursday, October 11th @ 3PM, location AA3195
by Guillaume Desjardins 11 Oct '12

11 Oct '12

A reminder for today's tea talk. The talk will be held in AA3195. ---------- Forwarded message ---------- From: Guillaume Desjardins <guillaume.desjardins(a)gmail.com> Date: Tue, Oct 9, 2012 at 3:23 PM Subject: Tea Talk: Thursday, October 11th @ 3PM, location TBD To: lisa LABO <lisa_labo(a)iro.umontreal.ca>, lisa_teatalk(a)iro.umontreal.ca Cc: fraser.maia(a)gmail.com Please join us this coming Thursday, for a tea-talk by Maia Fraser, who is visiting us from the University of Chicago. Title and abstract below. When: October 11th, 3pm Where: AA3195 Title: Structural Observations on Hierarchically Derived Reproducing Kernels Abstract: This talk will be informal. I will describe how convolution neural networks can be viewed within the mathematical framework proposed by [Smale2008] for hierarchically deriving reproducing kernels on spaces of images. I will then show how this axiomatic set-up makes it possible to use Hilbert space geometry in order to determine how certain design aspects (depth, type of neural response, choice of templates) can be traded off against each other. -- Guillaume

1 0

Tea Talk: Thursday, October 11th @ 3PM, location TBD
by Guillaume Desjardins 09 Oct '12

09 Oct '12

Please join us this coming Thursday, for a tea-talk by Maia Fraser, who is visiting us from the University of Chicago. Title and abstract below. When: October 11th, 3pm Where: TBD Title: Structural Observations on Hierarchically Derived Reproducing Kernels Abstract: This talk will be informal. I will describe how convolution neural networks can be viewed within the mathematical framework proposed by [Smale2008] for hierarchically deriving reproducing kernels on spaces of images. I will then show how this axiomatic set-up makes it possible to use Hilbert space geometry in order to determine how certain design aspects (depth, type of neural response, choice of templates) can be traded off against each other. -- Guillaume

1 0

Fwd: Help needed for user experiment
by Aaron Courville 04 Oct '12

04 Oct '12

---------- Forwarded message ---------- From: Joelle Pineau <jpineau(a)cs.mcgill.ca> Date: 2012/10/4 Subject: Help needed for user experiment To: aaron.courville(a)gmail.com Aaron, Can you forward the following to people in your group? thanks, Joelle ******************************** Chers collègues du LISA, Nous sommes à la recherche de quelques âmes généreuses pour participer à une étude sur le comportement des robots interactifs en situation d'interaction avec des humains. L'étude se tient dans les prochaines deux semaines, au pavillon Lassonde de l'Ecole Polytechnique de Montréal. Votre participation requiert une disponibilité de 30-45min., sur rendez-vous avec le chercheur principal (Andrew Sutcliffe). Vous aurez par le même coup l'occasion de voir en action le SmartWheeler, un fauteuil roulant intelligent, développé en collaboration entre le labo RLLab de McGill (sous la direction de Dr. Joelle Pineau) et le labo GRP de Poly (sous la direction de Dr. Richard Gourdeau). Svp contacter Andrew directement (sutcliffea(a)gmail.com) si vous êtes disponible. Merci! Joelle Pineau Associate Professor School of Computer Science McGill University, Montreal, Canada Tel: 514-398-5432 Fax: 514-398-3883 http://www.cs.mcgill.ca/~jpineau -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0

Tea Talk: Thursday, October 4th (tomorrow!) @ 3PM
by Guillaume Desjardins 04 Oct '12

04 Oct '12

Please join us tomorrow afternoon for Fred and Steven's presentation on code-review practices at LISA. When: October 4th, 3pm Where: LISA lab 1 (AA3256) Frédéric Bastien and Steven Pigeon will be presenting the new code review strategy to be used in the lab. The goal is to develop (and *use*) a framework in which the students can produce better code (especially when our commercial partners such as Ubisoft are concerned). We will discuss Assembla+Git, the proposed branching model, commit comments, and the git/perforce bridge with Ubisoft. While the discussion will be somewhat Ubi-centric, we strongly encourage all to attend as the code review process will also be used for other projects in the lab. -- Guillaume

2 3

TEA TALK TODAY!!!
by Aaron Courville 27 Sep '12

27 Sep '12

Greeting, We are restarting the tea talk series -- after the summer break -- with a good one. Our own Steven Pigeon will regale us with stories of BIG DATA. Apologizes for the *very* late notice (original email was blocked), but please make an effort to attend. When: TODAY, September 27th, 3pm Where: LISA lab 1 (AA3256) Title and abstract: MaDaMS: Massive Data Mining Summer School In this short presentation, we will have a look at what people in Data Mining have been up to, and how it relates and compares to what we are doing in machine learning. We will also discuss their perception of machine learning and why/how we should use this to our advantage. Cheers, Aaron -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0

TEA TALK TODAY AT 3PM
by Aaron Courville 27 Sep '12

27 Sep '12

Greeting, We are restarting the tea talk series -- after the summer break -- with a good one. Our own Steven Pigeon will regale us with stories of BIG DATA. Apologizes for the very late notice (entirely my fault), but please make an effort to attend. When: TODAY, September 27th, 3pm Where: LISA lab 1 (AA3256) Title and abstract: MaDaMS: Massive Data Mining Summer School In this short presentation, we will have a look at what people in Data Mining have been up to, and how it relates and compares to what we are doing in machine learning. We will also discuss their perception of machine learning and why/how we should use this to our advantage. Cheers, Aaron -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0

Tea Talk: ICML practice talk TODAY!
by Aaron Courville 21 Jui '12

21 Jui '12

Just to remind everyone of the ICML practice talk today at 15h00 (snacks provided) and to confirm that the location will indeed be AA3195. Cheers, Aaron On Wed, Jun 20, 2012 at 6:12 AM, Aaron Courville <aaron.courville(a)gmail.com> wrote: > Hi all, > > In our normal tea talk time slot, we will have an ICML practice talk > by Nicolas Boulanger-Lewandowski. Please be there with your > constructively critical hats on. > > When: 15h00 Thursday June 21, 2012 (i.e. tomorrow) > Where: TBA (hopefully AA3195) > > Title and Abstract: > Modeling Temporal Dependencies in High-Dimensional Sequences: > Application to Polyphonic Music Generation and Transcription > > We investigate the problem of modeling symbolic sequences of > polyphonic music in a completely general piano-roll representation. We > introduce a probabilistic model based on distribution estimators > conditioned on a recurrent neural network that is able to discover > temporal dependencies in high-dimensional sequences. Our approach > outperforms many traditional models of polyphonic music on a variety > of realistic datasets. We show how our musical language model can > serve as a symbolic prior to improve the accuracy of polyphonic > transcription. > > > Cheers, > Aaron > > -- > Aaron C. Courville > Département d’Informatique et > de recherche opérationnelle > Université de Montréal > email:Aaron.Courville@gmail.com -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0

Tea Talk: ICML practice talk
by Aaron Courville 20 Jui '12

20 Jui '12

Hi all, In our normal tea talk time slot, we will have an ICML practice talk by Nicolas Boulanger-Lewandowski. Please be there with your constructively critical hats on. When: 15h00 Thursday June 21, 2012 (i.e. tomorrow) Where: TBA (hopefully AA3195) Title and Abstract: Modeling Temporal Dependencies in High-Dimensional Sequences: Application to Polyphonic Music Generation and Transcription We investigate the problem of modeling symbolic sequences of polyphonic music in a completely general piano-roll representation. We introduce a probabilistic model based on distribution estimators conditioned on a recurrent neural network that is able to discover temporal dependencies in high-dimensional sequences. Our approach outperforms many traditional models of polyphonic music on a variety of realistic datasets. We show how our musical language model can serve as a symbolic prior to improve the accuracy of polyphonic transcription. Cheers, Aaron -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0

Reminder: Tea Talk Today!
by Aaron Courville 15 Jui '12

15 Jui '12

Details below: ---------- Forwarded message ---------- From: Aaron Courville <aaron.courville(a)gmail.com> Date: Wed, Jun 13, 2012 at 2:33 PM Subject: Tea Talk, new date and location! To: Lisa Labo <lisa_labo(a)iro.umontreal.ca>, lisa_teatalk(a)iro.umontreal.ca This week we have Stephan Gouws (http://ml.sun.ac.za/people/stephan-gouws) telling us about some of his work in NLP. Date/Time: Friday June 15, 2h00 Location: AA1177 Title: "Revisiting word representations for NLP" Cheers, Aaron -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0

Tea Talk, new date and location!
by Aaron Courville 13 Jui '12

13 Jui '12

This week we have Stephan Gouws (http://ml.sun.ac.za/people/stephan-gouws) telling us about some of his work in NLP. Date/Time: Friday June 15, 2h00 Location: AA1177 Title: "Revisiting word representations for NLP" Cheers, Aaron -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0

Tea Talk this Friday!
by Aaron Courville 13 Jui '12

13 Jui '12

This week we have Stephan Gouws (http://ml.sun.ac.za/people/stephan-gouws) telling us about some of his work in NLP. Date/Time: Friday June 15, 2h00 Location: TBA Title: "Revisiting word representations for NLP" Cheers, Aaron -- Aaron C. Courville Département d’Informatique et de recherche opérationnelle Université de Montréal email:Aaron.Courville@gmail.com

1 0