Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Reading List and Presentation Schedule

(CS 636 – Topics in Data Mining Research – Fall 2019-2020)

1. Introduction ful for sentence matching that has applications in ques-


tion answering, dialogue systems, and information re-
This document introduces the topics and suggested trieval. Nowadays, machine translation is dependent
reading for the seminar course CS 636 – Topics in Data upon deep neural network models for state-of-the-art
Mining Research offered in Fall 2019-2020 at LUMS. It performances. Deep architectures are effective for ro-
also gives the presentation schedule during the course. bust feature extraction and language modeling as well.
These models have great potential for processing of
2. Topics under-resourced languages like Urdu and Romanized
Urdu.
This year CS 636 will focus broadly on three top-
ics: (a) Assistive technologies and their applications. 2.3. Intelligent Crowd Sourcing
(b) Deep models for text representation and natural lan-
guage understanding, embeddings, and their application For any supervised learning tasks, annotated
for cross lingual text processing. (c) intelligent crowd dataset is required. These annotations are done man-
sourcing, quality assessment of crowd source, and cost ually by human annotators, which incur monetary cost
effective crowd sourcing. and is prone towards errors. An intelligent mechanism
to annotate the data can reduce the cost and mitigate er-
2.1. Assisitve Technologies rors. A mechanism of ranking the annotators in crowd
source setting is also a topic of interest as it provides
Assistive technology for the disabled people is a re- more confidence with respect to annotations if these are
search field that is gaining increasing prominence owing done by a “high ranked” annotator.
to an explosion of new interest in it from disparate dis-
ciplines. The field has a very relevant social impact on 3. Presentation Schedule
our ever-increasing aging and blind populations. The
focus of this topic would be to utilize AI and abundance The presentation schedule is given in the table be-
of available data to enable people with disabilities to cir- low. Please note that some papers are uploaded on
cumvent inaccessible situations. This would help them LMS. Missing papers can be downloaded from Google
to live more independently and get things done in their scholar by following the reference provided.
everyday lives.
References
2.2. Deep Learning
[1] M. Weiss, M. Luck, R. Girgis, C. Pal, and J. Cohen, “A
In recent years, deep models have become popu- survey of mobile computing for the visually impaired,”
lar for text and natural language processing. Textual arXiv preprint arXiv:1811.10120, 2018.
information is semi-structured in nature with sequen- [2] N. Vesdapunt, K. Bellare, and N. Dalvi, “Crowdsourc-
tial order and strong syntactic and semantic underpin- ing algorithms for entity resolution,” Proceedings of the
VLDB Endowment, vol. 7, no. 12, pp. 1071–1082, 2014.
nings. With the ever-growing availability of data, data-
[3] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit,
driven deep models have proven effective for represen-
L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin,
tation and understanding of textual information. A key “Attention is all you need,” in Advances in neural infor-
development in this direction has been distributed vec- mation processing systems, 2017, pp. 5998–6008.
torial representation of text (word2vec and its variants) [4] A. Manzoor, S. Arooj, S. Zulfiqar, M. Parvez, S. Shahid,
that has proven effective in many text and NLP appli- and A. Karim, “Alap: Accessible latex based mathemat-
cations. Deep neural networks have also been success- ical document authoring and presentation,” in Proceed-
Table 1. CS 636 Presentation Schedule(Paper number assigned to groups can be found from refer-
ences.)
Lecture Date Group 1 Group 2
1 03-09-2019 Asim Karim Intro
2 05-09-2019 Asim Karim Overview of topics
- 10-09-2019 Ashura Holiday -
3 12-09-2019 Qurat-Ul-Ain Quraishi & Muhammad Hassan [1] Muhammad Ahmad & Naimat Ullah [2]
4 17-09-2019 Moiz Ahmad & Muhammad Noman [3] Usman Khalid Mian & Unais Quyyum [4]
5 19-09-2019 Azan Bin Zahid & Hamza Liaqat [5] Zainab Tariq & Sayyeda Rijab [6]
6 21-09-2019 Saad Ahmad Khan & Soban Ali [7] Muhammad Nauman Minhas[8]
7 24-09-2019 Syed Anas & Rauhaan Rizvi[9] Muhammad Ahmad & Naimat Ullah [10]
8 26-09-2019 Usman Khalid Mian & Unais Quyyum [11] Azan Bin Zahid & Hamza Liaqat [12]
9 1-10-2019 Zainab Tariq & Muhammad Nauman Minhas [13] Saad Ahmad Khan & Soban Ali[14]
10 3-10-2019 Muhammad Hassan [15] Moiz Ahmad & Muhammad Noman[16, 17]
11 8-10-2019 Syed Anas & Rauhaan Rizvi[18] Muhammad Ahmad & Naimat Ullah[19]
12 10-10-2019 Usman Khalid Mian & Unais Quyyum[20] Azan Bin Zahid & Hamza Liaqat[21]
13 15-10-2019 Zainab Tariq & Muhammad Nauman Minhas[22] Saad Ahmad Khan & Soban Ali[23]
14 17-10-2019 Muhammad Hassan[24] Moiz Ahmad & Muhammad Noman[25]
15 22-10-2019 Syed Anas & Rauhaan Rizvi[26] Muhammad Ahmad & Naimat Ullah[27]
16 24-10-2019 Usman Khalid Mian & Unais Quyyum[28] Azan Bin Zahid & Hamza Liaqat[29]
17 29-10-2019 Zainab Tariq & Muhammad Nauman Minhas[30] Saad Ahmad Khan & Soban Ali[31]
18 31-10-2019 Muhammad Hassan[32] Moiz Ahmad & Muhammad Noman[33]
19 5-11-2019 Syed Anas & Rauhaan Rizvi[34]
20 7-11-2019

ings of the 2019 CHI Conference on Human Factors in [11] O. Inel, K. Khamkham, T. Cristea, A. Dumitrache,
Computing Systems, 2019, p. 504. A. Rutjes, J. van der Ploeg, L. Romaszko, L. Aroyo,
[5] A. Bhowmick and S. M. Hazarika, “An insight into as- and R.-J. Sips, “Crowdtruth: Machine-human compu-
sistive technology for the visually impaired and blind tation framework for harnessing disagreement in gather-
people: State-of-the-art and future trends,” Journal on ing annotated data,” in International Semantic Web Con-
Multimodal User Interfaces, vol. 11, no. 2, pp. 149–172, ference, 2014, pp. 486–504.
2017. [12] N. B. Shah and D. Zhou, “Double or nothing: Mul-
[6] H. Kacorri, “Teachable machines for accessibility,” tiplicative incentive mechanisms for crowdsourcing,”
ACM SIGACCESS Accessibility and Computing, no. in Advances in neural information processing systems,
119, pp. 10–18, 2017. 2015, pp. 1–9.
[7] S. Wu, L. Reynolds, X. Li, and F. Guzmán, “Design and [13] R. A. Krishna, K. Hata, S. Chen, J. Kravitz, D. A.
evaluation of a social media writing support tool for peo- Shamma, L. Fei-Fei, and M. S. Bernstein, “Embracing
ple with dyslexia,” in Proceedings of the 2019 CHI Con- error to enable rapid crowdsourcing,” in Proceedings of
ference on Human Factors in Computing Systems, 2019, the 2016 CHI conference on human factors in computing
p. 516. systems, 2016, pp. 3167–3179.
[8] A. Guo, E. Kamar, J. W. Vaughan, H. Wallach, [14] Y. Fu, T. M. Hospedales, T. Xiang, J. Xiong, S. Gong,
and M. R. Morris, “Toward fairness in ai for people Y. Wang, and Y. Yao, “Robust subjective visual property
with disabilities: A research roadmap,” arXiv preprint prediction from crowdsourced pairwise labels,” IEEE
arXiv:1907.02227, 2019. transactions on pattern analysis and machine intelli-
[9] J. Yang, J. Fan, Z. Wei, G. Li, T. Liu, and X. Du, “Cost- gence, vol. 38, no. 3, pp. 563–577, 2015.
effective data annotation using game-based crowdsourc- [15] J. C. Chang, S. Amershi, and E. Kamar, “Revolt: Col-
ing,” Proceedings of the VLDB Endowment, vol. 12, laborative crowdsourcing for labeling machine learning
no. 1, pp. 57–70, 2018. datasets,” in Proceedings of the 2017 CHI Conference on
[10] A. Nottamkandath, J. Oosterman, D. Ceolin, G. K. D. Human Factors in Computing Systems, 2017, pp. 2334–
de Vries, and W. Fokkink, “Predicting quality of crowd- 2346.
sourced annotations using graph kernels,” in IFIP Inter- [16] T. Mikolov, I. Sutskever, K. Chen, G. S. Corrado,
national Conference on Trust Management, 2015, pp. and J. Dean, “Distributed representations of words and
134–148. phrases and their compositionality,” in Proceedings of
Neural Information Processing Systems (NIPS), 2013, N. Shazeer, A. Ku, and D. Tran, “Image transformer,”
pp. 3111–3119. arXiv preprint arXiv:1802.05751, 2018.
[17] Y. Goldberg and O. Levy, “word2vec explained: deriv- [31] M. Jaderberg, K. Simonyan, A. Zisserman et al., “Spa-
ing mikolov et al.’s negative-sampling word-embedding tial transformer networks,” in Advances in neural infor-
method,” arXiv, 2014. mation processing systems, 2015, pp. 2017–2025.
[18] J. Pennington, R. Socher, and C. Manning, “Glove: [32] Z. Dai, Z. Yang, Y. Yang, W. W. Cohen, J. Carbonell,
Global vectors for word representation,” in Proceedings Q. V. Le, and R. Salakhutdinov, “Transformer-xl: At-
of the Conference on Empirical Methods in Natural Lan- tentive language models beyond a fixed-length context,”
guage Processing (EMNLP), 2014, pp. 1532–1543. arXiv preprint arXiv:1901.02860, 2019.
[19] M.-R. Bouguelia, S. Nowaczyk, K. Santosh, and [33] M. Dehghani, S. Gouws, O. Vinyals, J. Uszkoreit,
A. Verikas, “Agreeing to disagree: Active learning and Ł. Kaiser, “Universal transformers,” arXiv preprint
with noisy labels without crowdsourcing,” International arXiv:1807.03819, 2018.
Journal of Machine Learning and Cybernetics, vol. 9, [34] M. E. Peters, M. Neumann, M. Iyyer, M. Gardner,
no. 8, pp. 1307–1319, 2018. C. Clark, K. Lee, and L. Zettlemoyer, “Deep contextu-
[20] S. Wan, Y. Lan, J. Guo, J. Xu, L. Pang, and X. Cheng, “A alized word representations,” in Proceedings of NAACL-
deep architecture for semantic matching with multiple HLT, 2018, pp. 2227–2237.
positional sentence representations.” in Proceedings of
AAAI, 2016, pp. 2835–2841.
[21] B. Hu, Z. Lu, H. Li, and Q. Chen, “Convolutional neu-
ral network architectures for matching natural language
sentences,” in Proceedings of Neural Information Pro-
cessing Systems (NIPS), 2014, pp. 2042–2050.
[22] T. Bosc and P. Vincent, “Auto-encoding dictionary def-
initions into consistent word embeddings,” in Proceed-
ings of the 2018 Conference on Empirical Methods in
Natural Language Processing, 2018, pp. 1522–1532.
[23] C. N. Dos Santos and M. Gatti, “Deep convolutional
neural networks for sentiment analysis of short texts.” in
Proceedings of Conference on Computational Linguis-
tics (COLING), 2014, pp. 69–78.
[24] I. Sutskever, O. Vinyals, and Q. V. Le, “Sequence to se-
quence learning with neural networks,” in Proceedings
of Neural Information Processing Systems (NIPS), 2014,
pp. 3104–3112.
[25] A. Piktus, N. B. Edizel, P. Bojanowski, E. Grave, R. Fer-
reira, and F. Silvestri, “Misspelling oblivious word em-
beddings,” in Proceedings of the 2019 Conference of the
North American Chapter of the Association for Com-
putational Linguistics: Human Language Technologies,
Volume 1 (Long and Short Papers), 2019, pp. 3226–
3234.
[26] J. Guo, Y. Fan, Q. Ai, and W. B. Croft, “A deep rele-
vance matching model for ad-hoc retrieval,” in Proceed-
ings of the ACM International on Conference on Infor-
mation and Knowledge Management. ACM, 2016, pp.
55–64.
[27] S. Rajeswar, S. Subramanian, F. Dutil, C. Pal, and
A. Courville, “Adversarial generation of natural lan-
guage,” arXiv preprint arXiv:1705.10929, 2017.
[28] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova,
“Bert: Pre-training of deep bidirectional trans-
formers for language understanding,” arXiv preprint
arXiv:1810.04805, 2018.
[29] Y. Liu, M. Ott, N. Goyal, J. Du, M. Joshi, D. Chen,
O. Levy, M. Lewis, L. Zettlemoyer, and V. Stoyanov,
“Roberta: A robustly optimized bert pretraining ap-
proach,” arXiv preprint arXiv:1907.11692, 2019.
[30] N. Parmar, A. Vaswani, J. Uszkoreit, Ł. Kaiser,

You might also like