For the upcoming TSD2019, the following outstanding and eminent keynote speakers with respectable expertise in the conference interest fields agreed to give their respective pieces of speech (in alphabetical order):

The keynote speakers are brought to you by Springer

See the next section below for details about the speeches (topics, abstracts).


Ryan Cotterell

Ryan Cotterell

Lecturer – Natural Language and Information Processing Research Group, Dept. of Computer Science and Technology, University of Cambridge, United Kingdom

Mitigating Gender Bias in Morphologically Rich Languages

Abstract:  Gender bias exists in corpora of all of the world’s languages: the bias is a function what people talk about, not of the grammar of a language. For this reason, data-driven systems in NLP that are trained on this data will inherit such bias. Evidence of bias can be found in all sorts of NLP technologies: word vectors, language models, coreference systems and even machine translation. Most of the research done to mitigate gender bias in natural language corpora, however, has focused solely on English. For instance, in an attempt to remove gender bias in English corpora, NLP practitioners often augment corpora by swapping gendered words: i.e., if "he is a smart doctor" appears, add the sentence "she is a smart doctor" to the corpus as well before training a model. The broader research question asked in this talk is the following: How can we mitigate gender bias in corpora from any of the world’s languages, not just in English? As an example, the simple swapping heuristic for English will not generalize to most of the world’s languages. Indeed, such a solution would not even apply to German, since it marks gender on both nouns and adjectives and requires gender agreement throughout a sentence. MIn the context of German, this task is far more complicated: mapping "er ist ein kluger Arzt" to "sie ist eine kluge Ärztin" requires more than simply swapping "er" with "sie" and "Arzt" with "Ärztin"—one also has to modify the article ("ein") and the adjective ("klug"). In this talk, we present a machine-learning solution to this problem: we develop a novel neural random field that generates such sentence-to-sentence transformations, enforcing agreement with respect to gender. We explain how to perform inference and morphological reinflection to generate such transformations without any labeled training examples. Empirically, we illustrate that the model manages to reduce gender bias in corpora without sacrificing grammaticality with a novel metric of gender bias. Additionally, we discuss concrete applications to coreference resolution and machine translation.

Biography:  Ryan Cotterell is a lecturer (≈assistant professor) in the Department of Computer Science and Technology at the University of Cambridge. He will receive his Ph.D. from the Johns Hopkins Computer Science department in the Spring of 2019, where he was affiliated with the Center for Language and Speech Processing. He was co-advised by Jason Eisner and David Yarowsky. He specializes in natural language processing and machine learning, mostly publishing at *ACL and EMNLP, where he has published over 40 papers; he has won best paper awards at ACL 2017 and EACL 2017 after having twice runnered-up (EMNLP 2015, NAACL 2016). Previously, he was a visiting Ph.D. student at the Center for Information and Language Processing at Ludwig-Maximilians-Universität München supported by a Fulbright Fellowship and a DAAD Research Grant under the supervision of Hinrich Schütze. His Ph.D. was supported by a Facebook Fellowship, and an NDSEG graduate fellowship.
Denis Jouvet

Denis Jouvet

Senior Researcher – Inria Nancy, France

Speech Processing and Prosody  PDF (7 MB)

Abstract:  The prosody of the speech signal conveys information over the linguistic content of the message: prosody structures the utterance, and also brings information on speaker’s attitude and speaker’s emotion. Duration of sounds, energy and fundamental frequency are the prosodic features. However, their automatic computation and usage are not obvious. Sound duration features are usually extracted from speech recognition results or from a force speech-text alignment. Although the resulting segmentation is usually acceptable on clean native speech data, performance degrades on noisy or not non-native speech. Many algorithms have been developed for computing the fundamental frequency, they lead to rather good performance on clean speech, but again, performance degrades in noisy conditions. However, in some applications, as for example in computer assisted language learning, the relevance of the prosodic features is critical; indeed, the quality of the diagnostic on the learner’s pronunciation will heavily depend on the precision and reliability of the estimated prosodic parameters.

The talk will consider the computation of prosodic features, show the limitations of automatic approaches, and discuss the problem of computing confidence measures on such features. Then the talk with discuss the role of prosodic features and how they can be handled for automatic processing in some tasks such as the detection of discourse particles, the characterization of emotions, the classification of sentence modalities, as well as in computer assisted language learning and in expressive speech synthesis.

Biography:  Denis Jouvet is a senior researcher at Inria Nancy (France) since 2009. Previously he was with France Telecom R&D labs in Lannion (France), where he was involved in, and then also managing, automatic speech recognition studies and the development of automatic speech recognition technology for interactive vocal services. He participated in the European projects SMADA on directory assistance, and DIVINES on intrinsic speech variabilities. In 2009, he joined the PAROLE team in the LORIA laboratory in Nancy, and he is currently the team leader of the MULTISPEECH team. He was and is still involved into collaborative projects. He is author or co-author of more than 150 publications. His current main interests include speech modeling, automatic speech recognition, automatic speech synthesis, stochastic modeling, deep learning, signal processing, prosodic features, non-native automatic speech recognition and computer assisted foreign language learning.
Bhiksha Raj

Bhiksha Raj

Professor – School of Computer Science, Carnegie Mellon University, United States

Adversarial Attacks On ML Systems  PDF (4.7 MB)

Abstract:  As neural network classifiers become increasingly successful at various tasks ranging from speech recognition and image classification to various natural language processing tasks and even recognizing malware, a second, somewhat disturbing discovery has also been made. It is possible to fool these systems with carefully crafted inputs that appear to the lay observer to be natural data, but cause the neural network to misclassify in random or even targeted ways.

In this talk we will discuss why such attacks are possible, and the problem of designing, identifying, and avoiding attacks by such crafted "adversarial" inputs.

Biography:  Bhiksha Raj is a professor in the school of Computer Science at Carnegie Mellon University. He is also affiliated with the department of Electrical and Computer Enigneering and the Machine Learning Department at CMU. Dr. Raj's research interests span automatic speech recognition, speech and audio processing, machine learning theory, privacy and security, particularly as applied to speech processing, and deep learning, and has authored over 200 papers and 25 papers on these topics. Dr. Raj is a fellow of the IEEE.
Aline Villavicencio

Aline Villavicencio

Professor – Department of Computer Science, University of Sheffield, United Kingdom

Multiword Expressions and Idiomaticity: How Much of the Sailing Has Been Plain?  PDF (17 MB)

Abstract:  Precise natural language understanding requires adequate treatments both of single words and of larger units. However, expressions like idioms, verb-particle constructions and compound nouns may display idiomaticity, and while a police car is a car used by the police, a loan shark is not a fish that can be borrowed. Therefore it is important to identify which words in a sentence form an expression, and whether an expression is idiomatic, as this determines if it can be interpreted from a combination of the meanings of their component words or not. In this talk I discuss advances on the identification and treatment of multiword expressions in texts, focusing in particular on techniques for modelling idiomaticity.

Biography:  Aline Villavicencio is affiliated to the School of Computer Science and Electronic Engineering, University of Essex (UK) and to the Institute of Informatics, Federal University of Rio Grande do Sul (Brazil). She received her PhD from the University of Cambridge (UK) in 2001, and held postdoc positions at the University of Cambridge and University of Essex (UK). She was a Visiting Scholar at the Massachusetts Institute of Technology (USA, 2011-2012 and 2014-2015), at the École Normale Supérieure (France, 2014), an Erasmus-Mundus Visting Scholar at Saarland University (Germany in 2012/2013) and at the University of Bath (UK, 2006-2009). She held a Research Fellowship from the Brazilian National Council for Scientific and Technological Development (Brazil, 2009-2017). She is a member of the editorial board of TACL and of JNLE, PC Co-Chair of CoNLL-2019, Area Chair for ACL-2019, NAACL 2018, COLING 2018, ACL-2017 and General co-chair for the 2018 International Conference on Computational Processing of Portuguese. She is also a member of the SIGLEX board and of the program committees of various ACL and AI conferences, and has co-chaired several ACL workshops on Cognitive Aspects of Computational Language Acquisition and on Multiword Expressions. Her research interests include lexical semantics, multilinguality, and cognitively motivated NLP, and has co-edited special issues and books dedicated to these topics.
Back to top...