Midwest
Speech and
Language Days
2015
May 7-8, 2015
Toyota Technological Institute at Chicago
6045 S. Kenwood Ave, Chicago, IL

Midwest Speech and Language Days (MSLD) is a 2-day meeting held at TTI-Chicago that continues and expands upon the tradition of Illinois Speech Day and the Midwest Computational Linguistics Colloquium. Presenters and attendees come from Midwest universities and research institutions. The goal of the workshop is to increase awareness of speech and language research going on in the region and to foster collaboration among sites.

Please contact the organizers for more information or to register. There is no registration fee, and breakfast and lunch will be provided.

Call for participation

Invited Speakers:

Li Deng, Microsoft Research
Kevin Knight, ISI, USC
Haizhou Li, I2R, Singapore


Organizers:

Mohit Bansal, TTI-Chicago
David Chiang, Notre Dame
Kevin Gimpel, TTI-Chicago
Karen Livescu, TTI-Chicago


Details:

Talks:   Talk slots are 20 minutes and include 17 minutes for the talk and 3 for questions.
Posters:   Posters can be at most 4 feet by 6 feet. Please see your poster number below (day 1 and day 2 are different poster sets) and put your poster up at the corresponding number tag on the 5th floor walls.
Parking:   A limited number of parking passes are available for TTIC's parking lot. Please contact the organizers to reserve one. Park in the building's parking lot and get your parking pass from the security guard in the lobby. There is also free parking in the commuter parking lot at 60th St and Stony Island Ave and free street parking on many streets near TTIC (just beware of "permit parking" and "street cleaning" signs!). Parking can be found on 61st Street (between Woodlawn Ave and Blackstone Ave), on Dorchester Street (between 60th and 61st Streets).
Accommodations:   We are providing a limited number of student lodging grants at La Quinta (4900A S. Lake Shore Drive Chicago, IL 60615), which also has a shuttle to UChicago. Please get in touch with the organizers asap if you would like to apply for this grant. For those interested in staying within close walking distance of TTIC, there is also the Hyatt Place Chicago-South. There are also many reasonable options in downtown Chicago, such as Hotel Felix.


Schedule:

Thursday, May 7

10.00-10.55:     Breakfast provided at TTIC + Registration + Poster Setup

10.55-11.00:     Opening remarks

11.00-12.00:     Keynote session (Session chair: Karen Livescu)
Deep Learning for Speech and Language at Microsoft: A Selected Overview. Li Deng (MSR)

12.00-2.00:     Lunch provided at TTIC + Poster Session

2.00-2:45:     Invited talk (Session chair: Sadaoki Furui)
Conversational agents - first baby steps towards artificial intelligence and cognition. Haizhou Li (I2R, Singapore)

2.45-3.25:     Talk Session 1 (Session chair: Mark Hasegawa-Johnson)
2.45-3.05:     A deep neural network-based selection function for concatenative resynthesis of noisy and degraded speech. Michael Mandel, Young Suk Cho, Yuxuan Wang (OSU)
3.05-3.25:     Linear and nonlinear CCA for feature learning in speech and language tasks. Weiran Wang, Raman Arora, Karen Livescu, Jeff Bilmes (TTIC, JHU, UWash)

3.25-3.40:     Break

3.40-4.40:     Talk Session 2 (Session chair: Alan C. L. Yu)
3.40-4.00:     Neural Compositional Models based on Paraphrases. John Wieting, Mohit Bansal, Kevin Gimpel, Karen Livescu (UIUC, TTIC)
4.00-4.20:     Kneser-Ney smoothing on expected counts. Hui Zhang, David Chiang (Facebook, Notre Dame)
4.20-4.40:     Values in Words: Using Language to Evaluate and Understand Personal Values. Ryan L. Boyd, Steven R. Wilson, James W. Pennebaker, Michal Kosinski, David J. Stillwell, and Rada Mihalcea (UMich)

4.40-4.55:     Break

4.55-6.15:     Talk Session 3 (Session chair: Mohit Bansal)
4.55-5.15:     Probing the Linguistic Strengths and Limitations of Unsupervised Grammar Induction. Yonatan Bisk, Julia Hockenmaier (UIUC)
5.15-5.35:     Segmental Conditional Random Fields with Deep Neural Networks as Acoustic Models for First-Pass Word Recognition. Yanzhang He and Eric Fosler-Lussier (OSU)
5.35-5.55:     Investigating variation in English vowel-to-vowel coarticulation in a longitudinal phonetic corpus. Alan C. L. Yu, Carissa Abrego-Collier, Jacob Phillips, Betsy Pillion, Daniel Chen (UChicago)
5.55-6.15:     Exploring the assumptions of language acquisition models. Christos Christodoulopoulos, Dan Roth, Cynthia Fisher (UIUC)

6.15-6.30:     Take down remaining Day 1 posters, hang up Day 2 posters

Day 1 Posters:
  1. Decoding Strategies and Neural Network Features for Segmental Conditional Random Fields. Hao Tang, Weiran Wang, Kevin Gimpel, Karen Livescu (TTIC)
  2. Discriminative Probabilistic Graphical Models for Dialog State Tracking with Parameter Tying and External Knowledge. Yi Ma, Eric Fosler-Lussier (OSU)
  3. Deep Multilingual Correlation for Improved Word Embeddings. Ang Lu, Weiran Wang, Mohit Bansal, Kevin Gimpel, and Karen Livescu (Tsinghua, TTIC)
  4. Profile-based Authorship Matching. Jonathan Dunn, Amin Rasooli, Shlomo Engelson Argamon (IIT)
  5. Documenting Endangered Languages by Developing Speech-Corpora and ASRs: Crow, Hidatsa, Mandan. Damir Cavar (Indiana), Malgorzata E. Cavar (Indiana), Wilhelm Meya (LLC), John Boyle (CSU Fresno)
  6. Acoustic Correlates for Perceived Effort Levels in Expressive Speech... and Beyond. Mary Pietrowicz, Mark Hasegawa-Johnson, Karrie Karahalios (UIUC)
  7. Towards Learning to Follow Navigational Instructions with Deep Learning. Hongyuan Mei, Mohit Bansal, Matthew Walter (TTIC)
  8. Bootstrapping and Active Learning of Pronunciation Dictionaries. Mark Hasegawa-Johnson (UIUC)
  9. Decision Tree Framework for Spatiotemporal Sequence Prediction. Taehwan Kim, Yisong Yue, Sarah Taylor, Iain Matthews, Barry Theobald (TTIC, Disney)
  10. The Effect of Cognitive Load on Tonal Coarticulation. Katie Franich (UChicago)
  11. Extensions to corpus-based discovery of semantic intensity scales. Chaitanya Shivade, Marie-Catherine de Marneffe, Eric Fosler-Lussier, Albert M. Lai (OSU)
  12. Quantifying handshape similarity: a theory-driven approach. Jon Keane (UChicago)
  13. Dependency Link Embeddings: Continuous Representations of Syntactic Substructures. Mohit Bansal (TTIC)
  14. Lifelong Machine Learning for Topic Modeling and Beyond. Zhiyuan Chen and Bing Liu (UIC)


Friday, May 8

9.00-10.00:     Breakfast provided at TTIC + Poster Session

10.00-11.00:     Keynote session (Session chair: David Chiang)
How Much Information Does a Human Translator Add to the Original? Kevin Knight (ISI, USC)

11.00-11.10:     Break

11.10-12.10:     Talk Session 4 (Session chair: Kevin Gimpel)
11.10-11.30:     Entity-Parse Semantics. David McAllester (TTIC)
11.30-11.50:     Acquiring Speech Transcriptions Using Mismatched Crowdsourcing. Preethi Jyothi and Mark Hasegawa-Johnson (UIUC)
11:50-12:10:     A Framework for Organizing and Advancing our Knowledge of Stuttering: The FLUSTER Model. Torrey M. Loucks, Anna Tendera, Sazzad Nasir (UIUC, Northwestern)

12.10-12.20:     Break

12.20-1.00:     Talk Session 5 (Session chair: Michael Mandel)
12:20-12:40:     Enabling Resources in Digital Language Archives using Automatic Speech Recognition. Damir Cavar, Malgorzata E. Cavar, Lwin Moe, Aaron Albin (Indiana)
12.40-1.00:     Learning Models for Following Natural Language Directions in Unknown Environments. Matthew R. Walter (TTIC)

1.00-1.05:     Closing remarks

1.05-2.30:     Lunch provided at TTIC + Poster Session

Day 2 Posters:
  1. Solving the paradigmatic alignment problem in unsupervised morphological paradigm induction. Jackson Lee and John Goldsmith (UChicago)
  2. Solving Hard Co-reference Instances. Daniel Khashabi, Haoruo Peng, Dan Roth (UIUC)
  3. Structural Learning with Amortized Inference. Kai-Wei Chang, Shyam Upadhyay, Gourab Kundu and Dan Roth (UIUC)
  4. PseudoMatic: a flexible pseudoword generator with triphones. Needle, J. M., Pierrehumbert, J. B., Hay, J. B. (Northwestern)
  5. Using Syntax and Semantics for Improved Reading Comprehension. Hai Wang, Mohit Bansal, Kevin Gimpel, David McAllester (TTIC)
  6. Generating Disambiguating Paraphrases for Use in Crowdsourced Judgments of Meaning. Ethan Hill, Manjuan Duan, Michael White (OSU)
  7. A Sense-Topic Model for Word Sense Induction with Unsupervised Data Enrichment. Jing Wang, Mohit Bansal, Kevin Gimpel, Brian Ziebart, and Clement Yu. (UIC, TTIC)
  8. Exploring spectrotemporal derivative features in DNN based speech recognition. Deblin Bagchi, Yanzhang He, Eric Fosler-Lussier (OSU)
  9. Generalization of Error Tokens of Hearing Impaired Consonant Perception. Cliston L. Cole, Jont B. Allen (UIUC)
  10. Translating Sound Adjectives with a Crowdsourced Concept Map. Mark Cartwright and Bryan Pardo (Northwestern)
  11. Gender Prediction for Chinese Social Media Data. Wen Li and Markus Dickinson (Indiana)
  12. Visual Feedback Facilitates Suppression of the Pitch-Shift Response. Li-Hsin Ning, Torrey Loucks and Chilin Shih (UIUC)
  13. Towards a highly customizable cochlear implant speech processing pipeline. Taher S. Mirzahasanloo, and Nasser Kehtarnavaz (UTDallas)
  14. How NALR tuned hearing aids help HI ears in speech perception. Ali Abavisani (UIUC)


Participants:
We had ⁓90 participants, from the following universities:
  1. Illinois Institute of Technology Chicago
  2. Indiana University Bloomington
  3. Northwestern University
  4. Ohio State University
  5. TTI-Chicago
  6. University of Chicago
  7. University of Illinois Urbana-Champaign
  8. University of Illinois Chicago
  9. University of Notre Dame
  10. University of Michigan Ann Arbor
  11. University of Texas at Dallas

Some previous workshops:
2014
2013
2012
2011