UPLINX-Logo

Semantic Deep Learning

Seminar an der Universitšt des Saarlandes, Fachrichtung Informatik, LSF 116956

The topics for the seminar time slots are listed below. Currently assigned topics are marked in red; waiting list for topics is maintained.   

Selected background literature is indicated below; topic reference papers are available in the web or on request from seminar organizers. In the following, “DL” abbreviates “Deep Learning”.

Topic#

Topic

1

DL-based link prediction in knowledge graphs for semantic scene descriptions:

S. Baier, Y. Ma, V. Tresp (2017): Improving Visual Relationship Detection using Semantic Modeling of Scene Descriptions. Proc. of International Semantic Web Conference (ISWC); LNCS, Springer.
https://arxiv.org/pdf/1809.00204.pdf  and https://arxiv.org/pdf/1808.08941.pdf  

2

DL-based translation from natural language to description logic (ALCQ) for ontology learning:

G. Petrucci, M. Rospocher, C. Ghidini (2018): Expressive Ontology Learning as Neural Machine Translation. Journal of Web Semantics; Elsevier.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3274562

3

DL-based natural language inference I

S. Bowman, C. Potts, CD. Manning (2015): Recursive Neural Networks Can Learn Logical Semantics. Proc. of 3rd Workshop on Continuous Vector Space Models and their Compositionality; ACL.
https://arxiv.org/pdf/1406.1827.pdf  and Tree-structured recursive neural networks for sentence meaning (recognition of textual entailment between sentences).

4

DL-based natural language inference II

(a) J. Devlin, M.-W. Chang, K. Lee,  K. Toutanova (2018):  Bert: Pre-Training of Deep Bidirectional Transformers for Language Understanding. arXiv preprint arXiv:1810.04805.
https://arxiv.org/pdf/1810.04805.pdf  [ref. in 10.]

(b) Q. Chen, X. Zhu, Z.-H. Ling, S. Wei, H. Jiang, D. Inkpen (2017): Enhanced LSTM for Natural Language Inference. Proc. of 55th Annual Meeting of the Association for Computational Linguistics; vol.1:1657–1668. https://arxiv.org/pdf/1609.06038.pdf  [ref. in 10.] 

5

Ontology-based deep learning of classification and prediction tasks:

(a) H. Wang, D. Dou, D. Lowd (2016): Ontology-Based Deep Restricted Boltzmann Machine.
Proc. International Conference on Database and Expert Systems Applications (DEXA); LNCS, Springer.

(b) N. Phan, D. Dou, H. Wang, D. Kil, B. Piniewski (2015): Ontology-Based Deep Learning for Human Behavior Prediction in Health Social Networks. Proc. of 6th ACM Conference on Bioinformatics, Computational Biology and Health Informatics; ACM.

6

Generation of image descriptions via DL-based alignment of visual and language data:

A. Karpathy, L. Fei-Fe (2015): Deep Visual-Semantic Alignments for Generating Image Descriptions.
Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR); IEEE.

7

Generation of justification explanations of visual evidence and DL results:

LA. Hendricks, Z. Akata, M. Rohrbach, J. Donahue, B. Schiele, T. Darrell (2016): Generating Visual Explanations. Proc. of European Conference on Computer Vision; Springer.
https://arxiv.org/pdf/1603.08507.pdf  

8

Generation of introspective explanations via DL-based visual attention and annotation:

J. Kim, A. Rohrbach, T. Darrell, J. Canny, Z. Akata (2018): Textual Explanations for Self-Driving Vehicles. Proc. of European Conference on Computer Vision; Springer.
https://arxiv.org/pdf/1807.11546.pdf  and https://arxiv.org/abs/1612.04757 

9

Rationalization of DL result with backpropagation-based input feature relevance (attribution) methods:

M. Ancona, E. Ceolini, C. Oztireli, M. Gross (2018): Towards better understanding of gradient-based attribution methods for Deep Neural Networks. Proc. of 6th International Conference on Learning Representations (ICLR).

10

Visual commonsense reasoning:

R. Zellers, Y. Bisk, A. Farhadi, Y. Choi (2018): From Recognition to Cognition: Visual Commonsense Reasoning. arXiv:1811.10830. https://arxiv.org/pdf/1811.10830.pdf [ref. 4a+b]

 

Background:

Impressum