Seminar "Selected Topics in Human Language Technology and Pattern Recognition"

In the summer semester 2015 the Lehrstuhl Informatik 6 will host a seminar entitled "Selected Topics in Human Language Technology and Pattern Recognition".

Registration for the seminar

Registration for the seminar is only possible online via the registration page provided by the Computer Science Department.

Prerequisites for participation in the seminar

Seminar format and important dates

As discussed during the kick-off meeting, the seminar presentations will take place during the lecture period, always on Mondays at 14:00-16:00h. The start date of the presentations will be June 8, 2015. Below in the list of topics you find a first draft for the schedule for each of the presentations.

Note: failure to meet deadlines, absence without permission from compulsory sessions (presentations and preliminary meeting as announced by email to each participating student), or dropping out of the seminar after more than 3 weeks after the kick-off meeting (i.e. after March 23, 2015) results in the grade 5.0/not appeared.

Topics, relevant references and participants

The specific topics of the seminar will be can be found below and are introduced and distributed during the kick-off meeting in the seminar room 6010 in the ground floor of the Lehrstuhl Informatik 6.

The following main references will build the basis for this seminar:
  1. D. Yu, L. Deng: Automatic Speech Recognition: A Deep Learning Approach, Springer-Verlag, London 2015, 321 pages.
  2. X. He, L. Deng: Discriminative Learning for Speech Recognition: Theory and Practice, Morgan and Claypool Publishers, 2008, 112 pages.
  3. A. Graves, S. Fernandez, F. Gomez, J. Schmidhuber: "Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural Networks," in Proc. 23rd Intern. Conf. on Machine Learning (ICML), pp. 369-376, Pittsburgh, PA, 2006.
  4. H. Schwenk: "Continuous Space Language Models," in Computer Speech & Language, Vol. 21, No. 3, pp. 492-518, 2007.
  5. T. Mikolov, M. Karafiát, L. Burget, J. Cernocký, S. Khudanpur: "Recurrent Neural Network based Language Model," in Proc. Interspeech 2010, pp. 1045-1048, Makuhari, Chiba, Japan, Sept. 2010.
  6. T. Mikolov, S. Kombrink, L. Burget, J. Cernocký, S. Khudanpur: "Extensions of Recurrent Neural Network Language Model," in Proc. IEEE Intern. Conf. on Acoustics, Speech and Signal Processing (ICASSP), pp. 5528-5531, Prague, Czech Republic, May 2011.

Topics, References, Participants, and Supervisors:
  1. Automatic Speech Recognition: Introduction and Applications [1, Sec. 1] (NN; Betreuer: Christian Oberdörfer)
  2. Gaussian Mixture HMMs [1, Secs. 2 & 3] (NN; Betreuer: Christian Oberdörfer)
  3. Discriminative Training of Generative Models [2] (NN; Betreuer: Ralf Schlüter)
  4. Deep Neural Networks [1, Sec. 4] (Chebbi; Betreuer: Tamer Alkhouli)
    Vortrag: 08.06.2015
  5. Advanced Model Initialization [1, Sec. 5] (Sun; Betreuer: Tamer Alkhouli)
    Vortrag: 22.06.2015
  6. DNN Hybrid Modeling [1, Sec. 6] (Gleim; Betreuer: Eugen Beck)
    Vortrag: 15.06.2015
  7. Training and Decoding Speedup [1, Sec. 7] (NN; Betreuer: Eugen Beck)
  8. DNN Sequence-Discriminative Training [1, Sec. 8] (NN; Betreuer: Albert Zeyer)
  9. CTC-Training [3] (Ellers; Betreuer: Albert Zeyer)
    Vortrag: 15.06.2015
  10. Feature Representation Learning in DNNs [1, Sec. 9] (NN; Betreuer: Zoltan Tüske)
    Vortrag: 22.06.2015
  11. Tandem Modeling [1, Sec. 10] (Stanchev; Betreuer: Zoltan Tüske)
    Vortrag: 22.06.2015
  12. DNN Adaptation [1, Sec. 11] (Rogner; Betreuer: Pavel Golik)
    Vortrag: 29.06.2015
  13. Representation Sharing and Transfer in DNNs [1, Sec. 12] (Niu; Betreuer: Pavel Golik)
    Vortrag: 29.06.2015
  14. Recurrent and Convolutional Neural Networks [1, Secs. 13 and 15.1.7] (Wang; Betreuer: Harald Hanselmann)
    Vortrag: 29.06.2015
  15. Computational Networks [1, Sec. 14] (Boldbaatar; Betreuer: Harald Hanselmann)
    Vortrag: 06.07.2015
  16. Feed-Forward Neural Network-based Language Modeling [4] (Schwaiger; Betreuer: Kazuki Irie)
    Vortrag: 06.07.2015
  17. Recurrent Neural Network-based Language Modeling [5] and [6] (Zhang; Betreuer: Kazuki Irie)
    Vortrag: 06.07.2015

Guidelines for the article and presentation

The roughly 20-page article together with the slides (between 20 & 30) for the presentation should be prepared in LaTeX format. Presentations will consist of 40 minutes presentation time & 15 minutes discussion time. Document templates for both the article and the presentation slides are provided below along with links to LaTeX documentation available online. The article and the slides have to be prepared in LaTeX format using the provided templates and submitted electronically in pdf format. Other formats will not be accepted.




Inquiries should be directed to the respective supervisors or to:

Dr. Ralf Schlüter
RWTH Aachen
Lehrstuhl Informatik 6
Ahornstr. 55
52056 Aachen Raum 6125b
Telefon: 0241 / 80-21612