2017

Écouter la conférence du 10 novembre 2017

Conférencier : Stan Matwin, Ph. D., professeur et titulaire de la Chaire de recherche du Canada (niveau 1) à Dalhousie University, Halifax, NE, directeur de l'Institute  for Big Data Analytics, professeur émérite à l’Université d’Ottawa, et professeur titulaire à l’Institut d’informatique de l’Académie polonaise des sciences. Ses intérêts de recherche portent sur l'apprentissage automatique, en particulier à partir des données textuelles, ainsi que sur les applications d’exploration des données et sur la  privauté des données.

Résumé : Nous allons en premier lieu décrire les grandes lignes directrices des recherches menées dans l’Institut de Big Data Analytics à Dalhousie, en particulier nos travaux avec les données spatio-temporelles sur l'exploitation des océans. Nous allons ensuite faire un tour d’horizon d’exploration des données textuelles (EDT, ou text mining), un sous-domaine d’exploration de données tout court (data mining), en commençant par les méthodes vectorielles  classiques de représentation des textes, ainsi que les tâches les plus communes en EDT (clustering, classification, extraction d’information, et l’analyse du sentiment). Nous allons présenter à vol d’oiseau les méthodes modernes pour certaines de ces tâches, en particulier Conditional Random Fields et Latent Dirichlet Analysis. Dans la troisième partie de la présentation, nous allons introduire les nouvelles méthodes de représentation des données textuelles, fondées sur deep learning et la contextualisation (embeddings) des mots [Mikolov 2013]. Nous allons discuter les avantages de ces méthodes et leurs effet sur les tâches introduites au départ.

 

Conférencier invité :

Andrew D. Cohen, Professor Emeritus, University of Minnesota

Jeudi 25 mai 2017, 15h00
Université du Québec à Montréal
Pavillon J.-A. DeSève
Salle DS-1950
320, rue Sainte-Catherine Est, Montréal
Métro Berri-UQAM

Résumé :

This plenary will highlight a series of areas deemed worthy of attention by language researchers. In some cases the research effort would entail following up on studies initiated some years ago and in other cases the effort would involve relatively new research thrusts. The talk will include ideas about research in the following areas:

Pathways to success in language learning

• Language learners as informed consumers - the impact of coaching language learners in being more informed consumers of target language (TL) programs.
• The role of motivation in the L2-FL interface - the reality behind the second vs. foreign language (L2-FL) distinction as it applies to learners at differing levels of motivation.
• The language of thought for learning the target language - the differential effect of using the first language (L1) or the TL as the language of thought.
• The impact of L2/FL error correction over time,
• The use and impact of websites accessed in support of language learning,
• Language attrition over time - the nature of language attrition over time for people at different age and proficiency levels.

A close look at language learner strategies

• The fluctuating functions of strategies - the fluctuation in functions of language learner strategies (LLS).
• Refining strategies for language learning - research on specialized strategy areas such as grammar strategies and vocabulary learning strategies,
• The language strategies of hyperpolyglots - the strategies of hyperpolygots in their learning of 12 or more languages.
• Test-taking strategies - the strategies that respondents actually use in order to produce responses on tests of summarizing, cloze, and other response formats.
• Pushing the envelope with regard to TL pragmatics
• The less researched speech acts - descriptive studies of speech acts yet to be researched extensively.
• The effects of explicit instruction in pragmatics,
• The learning of pragmatics in the TL classroom from native- and non-native teachers,
• The teaching of pragmatics in World Englishes – the actual curriculum.

RSVP à isc@uqam.ca avant le 23 mai.

Activité conjointe de l’Institut des sciences cognitives et de l’École de langues.

Conférencière invitée :

Reyes Llopis-García (http://laic.columbia.edu/author/1501198000/) Columbia University, Department of Latin American and Iberian Cultures.

Vendredi 21 avril 2017, 14h30
Université du Québec à Montréal
Pavillon J.-A. DeSève
Salle DS-M425
320, rue Sainte-Catherine Est, Montréal
Métro Berri-UQAM

Résumé :

Cognitive Linguistics provides a theoretical framework to understand language as a builder of meaning that stems from interaction with the world and draws from the sensorimotor experience of all speakers. This point of view affords remarkable potential for foreign language teaching: it shows that as physical experiences are universal, so are the conceptualizations of the L1/L2 speakers who have them.

Inscription avant le 19 avril, à isc@uqam.ca.

Information

Conférencier invité :

Roger Levy, Associate Professor in the Department of Brain and Cognitive Sciences at the Massachusetts Institute of Technology

Vendredi 7 avril 2017, 15h00
Université du Québec à Montréal
Pavillon J.-A. DeSève
Salle DS-1950
320, rue Sainte-Catherine Est, Montréal
Métro Berri-UQAM

Résumé :

The infinite generative potential of human language derives from our ability to analyze complex linguistic input into simpler units, store those units in memory, and productively recombine those units into new expressions. This is the cycle of comprehension, acquisition, and production through which human languages persist and change through the history of a speech community. But what are these units of comprehension, acquisition, and production? The tension between combinatorial and holistic representation of complex linguistic expressions plays a central role in debates on language processing and acquisition. Here I describe work combining probabilistic models and new large datasets to investigate this tension and uncover the respective contributions of productive knowledge and direct experience. In processing, we focus on binomial expressions (salt and pepper - pepper and salt), finding a frequency-driven tradeoff between the two knowledge sources and a frequency-dependent level of idiosyncrasy in binomial ordering preference across binomials in the language. The former is explained by a rational model of learning from limited experience; the latter we account for with an evolutionary model of transmission of ordering preferences over time. In acquisition, we focus on determiner-noun combinations (“the ball”, “a cold”) and develop a novel Bayesian model to infer the strength of contribution of productive knowledge evident in child speech. We find evidence of low initial levels of productivity and higher levels later in development, consistent with the hypothesis that the earliest months of multi-word speech are not generated using rich grammatical knowledge, but that grammatical productivity emerges rapidly thereafter.

 

RSVP à isc@uqam.ca avant le 6 avril.

Conférence NeuroQAM, en collaboration avec l'Institut des sciences cognitives

Conférencière invitée :

Maithilee Kunda (https://my.vanderbilt.edu/mkunda/), Department of Electrical Engineering and Computer Science, Vanderbilt University.

Jeudi 6 avril 2017, 15h00
Université du Québec à Montréal
Local SU-1550
Pavillon Adrien-Pinard (SU)
100, rue Sherbrooke Ouest, Montréal

Résumé :

Despite evidence for the importance of visual mental imagery from many of the cognitive sciences, the field of artificial intelligence (AI) has not yet provided a rigorous computational account of how mental imagery works. Part of the problem comes from confusion in the AI literature between tasks that are presented visually, in an external visual format, versus tasks that are solved visually, using internal visual representations. There is a rich history of AI research in the first category, but the vast majority of these AI systems first convert visual inputs into internal propositional (i.e. abstract/symbolic) representations before solving a task. Fewer systems fall into the second category, but these studies have begun to provide insight into the computational nature of mental imagery and its role in intelligent behavior. I will present a synthesis of AI research into mental imagery over the years, including my recent work in developing AI systems that use purely visual representations to solve problems from standardized intelligence tests, as well as what these systems can tell us about visual mental imagery in typical development and in cognitive conditions such as autism.

ATELIER PRATIQUE: Construire un algorithme de raisonnement physique pour comparer le raisonnement des humains vs des systèmes d'intelligence artificielle. Par Maithilee Kunda (Département de génie électrique et informatique, Vanderbuilt University) Hands-on workshop: Comparing physical reasoning in humans and in AI system

How do people mentally represent their knowledge, for example about how objects physically move in the world?  Artificial intelligence (AI) systems give us a way of explicitly constructing different types of knowledge representations, and then conducting experiments to see how each representation supports capabilities related to learning and intelligent behavior.  In this workshop, we will explore the construction of knowledge representations that have to do with physical reasoning---how physical objects move, tip over, and roll.  Specifically, we will examine how physical knowledge can be represented using visual mental images and simple imagery-based reasoning operations.  We will implement a basic AI system that can reason about pictures of 2D blocks and shapes, and compare results to what humans can do.  We will be using Matlab; students with any level of programming experience are welcome (including those with no prior experience).

Vendredi 7 avril 9h30-12h30 au laboratoire informatique PK-S1555 (Atelier NeuroQAM)

Aucune expérience de programmation requise, venez essayer!

Un événement organisé par l'Institut des sciences cognitives et le centre NeuroQAM

Inscriptions requises avant le 5 avril 2017. SVP m'indiquer dans le courriel votre niveau de connaissances en programmation : aucune expérience, débutant, intermédiaire, avancé.

terraza.jimena@uqam.ca

Cette conférence est organisée conjointement avec le Centre de recherche sur le langage, l'esprit et le cerveau (CRLEC-UQAM)

Conférencière :

Dr. Denise Klein, Director, Centre for Research on Brain, Language and Music.  Assistant Professor, Neurology and Neurosurgery, Faculty of Medicine; Neuropsychologist, Cognitive Neuroscience Unit, Montreal Neurological Institut, McGill University

Vendredi 24 mars 2017, 15h00
Université du Québec à Montréal
Pavillon J.-A. DeSève
Salle DS-1950
320, rue Sainte-Catherine Est, Montréal
Métro Berri-UQAM

Résumé :

How does our early experience with language impact the human brain, higher cognitive functions, and learning? In this presentation, I will focus on research that we have conducted spanning three decades in which we combined behavioral methods with functional neuroimaging (PET and fMRI) to investigate how neural recruitment is influenced by the age of acquisition/exposure (AoE), proficiency in the language, and the distinctive characteristics of languages. I will discuss the use of relatively new anatomical techniques such as voxel-based morphometry-VBM, cortical thickness measures, diffusion tensor tractography to enhance our understanding of the critical-period phenomena and neural plasticity in the human brain. The program of research addresses the extent to which the human brain has the capacity to change as a result of learning. Here, we specifically investigate the extent to which the neural patterns are fixed and the extent to which the patterns can be altered later in life. The results of these studies reveal the neural underpinnings of human brain development in relation to the age of language exposure, and they suggest periods when learning language are most optimal in early life.

RSVP à isc@uqam.ca avant le 22 mars.

Conférencier :

Laurent Gosselin, Professeur de linguistique française à l'Université de Rouen-Normandie

Vendredi 27 janvier 2017, 15h00
Université du Québec à Montréal
Pavillon J.-A. DeSève
Salle DS-1950
320, rue Sainte-Catherine Est, Montréal
Métro Berri-UQAM

Résumé : 

En sémantique des formes verbales, la structure modale du temps est généralement conçue comme asymétrique, opposant le passé irrévocable à l’avenir ouvert, simplement possible – le moment présent opérant la « coupure modale ». La question est de savoir quelle est la nature de ce temps asymétrique. S’agit-il du temps métaphysique, objectif (position couramment admise en sémantique vériconditionnelle), ou du temps subjectif (position des sémantiques énonciativiste et cognitive) ? On défendra la thèse selon laquelle il s’agit d’un troisième type de temporalité, le temps pragmatique, et que cette asymétrie est liée au processus d’ajustement entre le monde et l’énoncé. On avancera aussi l’idée qu’il faut lui adjoindre un temps sémantique, également asymétrique, mais faisant passer la coupure modale au moment choisi comme moment de référence ; d’où le rôle central des visées aspectuelles. C’est de la confrontation, convergente ou divergente, de ces diverses temporalités que résultent, de façon régulière et prédictibles, les effets de sens modaux des formes verbales. On analysera plus particulièrement les valeurs du futur simple en français.

RSVP à isc@uqam.ca avant le 25 janvier.