AI helps you reading Science

AI generates interpretation videos

AI extracts and analyses the key points of the paper to generate videos automatically


pub
Go Generating

AI Traceability

AI parses the academic lineage of this thesis


Master Reading Tree
Generate MRT

AI Insight

AI extracts a summary of this paper


Weibo:
To motivate the need for our contributions, we present a representative

Seq2Seq-Vis: A Visual Debugging Tool for Sequence-to-Sequence Models.

IEEE Transactions on Visualization and Computer Graphics, no. 1 (2019): 353-363

Cited: 144|Views313
EI WOS

Abstract

Neural sequence-to-sequence models have proven to be accurate and robust for many sequence prediction tasks, and have become the standard approach for automatic translation of text. The models work with a five-stage blackbox pipeline that begins with encoding a source sequence to a vector space and then decoding out to a new target sequen...More

Code:

Data:

0
Introduction
  • Deep learning approaches based on neural networks have shown signif- the source sequence.
  • With enough data, these models provide a general icant performance improvements on many artificial intelligence tasks.
  • The complex structure of these networks often makes it dif- While the impact of seq2seq models has been clear, the added comficult to provide explanations for their predictions.
  • The high-dimensional internal represenshown state-of-the-art performance in a broad range of applications tations make it difficult to analyze the model as it transforms the data
Highlights
  • Deep learning approaches based on neural networks have shown signif- the source sequence
  • To motivate the need for our contributions, we present a representative
  • Seq2Seq-Vis is the result of an iterative design process and discuswhen using rule-based techniques, a user can explore the provenance und so haben wir entdeckt , dass es eine unendliche an gehäkelten hyperbolischen wesen gibt
  • In regular of a decision through rules activated for a given output
  • For vectors with many connections, we reduce visual clutter by computing a concave hull for all related neighbors and highlight the related dots within the hull
  • We considered several different variants for both main views of the system
Methods
  • DESIGN OF

    Seq2Seq-Vis less problematic in previous generations of AI systems. For instance als erstes muss man beachten , dass es gegenden auf dieser welt gibt , die wegen mangelnder aufmerksamkeit im dunkeln stehen .

    Seq2Seq-Vis is the result of an iterative design process and discuswhen using rule-based techniques, a user can explore the provenance und so haben wir entdeckt , dass es eine unendliche an gehäkelten hyperbolischen wesen gibt .

    sions between experts in machine learning and visualization.
  • Seq2Seq-Vis is the result of an iterative design process and discuswhen using rule-based techniques, a user can explore the provenance und so haben wir entdeckt , dass es eine unendliche an gehäkelten hyperbolischen wesen gibt.
  • Meetings the authors evaluated a series of low-fidelity prototypes and tested mistake in the system, an analyst can 1) identify which rule misfired, 2) them wir vergrößern das blickfeld , wir zoomen raus , durch eine nukleare pore , welche der zugang zu dem teil , der die dna beherbergt , ist und nukleus genannt wird .
Conclusion
  • Seq2Seq-Vis is a tool to facilitate deep exploration of all stages of a seq2seq model.
  • The authors apply our set of goals to deep learning models that are traditionally difficult to interpret.
  • Being an open source project, the authors see future work in evaluating the longitudinal feedback from real-world users for suggested improvements.
  • The authors already observed some initial quantitative and qualitative feedback.
  • More 5,500 page views have been recorded and 156 users liked the project on Github.
  • The most requested new feature is integration of the tool with other ML frameworks
Related work
  • Bemerkungen gesprochen hat , umzusetzen . On the face of it these two proposals appear to b

    Various methods [4, 34]

    have been proposed to generate explanations procedural changes to facilitate freedom of movement give e ect to the recent court cases Mrs Ber@@ ger ref for deep learning model predictions. Understandingretmhaerkms . ll remains a difficult task. To better address the specific issuZewseiteonfs eorwuährntue esredriesN,ach@@ wahlen . we narrow the target audience for our proposed tool. Following the

    Secondly , he also mentioned their by @-@ ele classifications by Strobelt et al [48] and Hohman et al [13], our tool aims at model developers who have at least a conceptuGaelsteurnn -deeinrigsetvaonndIhinnenghaben das bereits komm@ Wirtscha s@@ ausschuß von der Verbesserung der wir of how the model works. This is opposed to end userisn,Ewurohpoa . are agnostic to the technique used to arrive at a specific result. FolloiYnesgterHdaoy ,hams soamne eoftyou have noted , the Com al., analysis itself can broadly be divided into global Mmoonedtaeryl Aanaiarslryesfeirsredantodthe improvement in the e instance-based analysis. In global model analysis, thEuerompeo. commonly seen methods are visualizations of the internal struDcetruAbrgeeorodnfettersapriancheaduch von der modernenTec deep learning models. Instance-based analysis may eThceohuonpolueradblewMeimthber also mentioned mode interactive experimentation with the goal of understanDdieinKogmmaispsaarirntsipcrauclhavron der Bedeutung nationa prediction using the local information around only oStnaaetenininpduertKo[m3m8u]n.ikations@@ politik .
Funding
  • Demonstrates the utility of our tool through several real-world sequence-to-sequence use cases on large-scale models
  • Proposes SEQ2SEQ-VIS , a visual analytics tool that satisfies this criteria by providing support for the following three goals: M
  • Presents a guiding example illustrating how a typical model understanding- and debugging session looks like for an analyst
Reference
  • D. Alvarez-Melis and T. S. Jaakkola. A causal framework for explaining the predictions of black-box sequence-to-sequence models. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, EMNLP 2017, Copenhagen, Denmark, September 9-11, 2017, pp. 412–421, 2017.
    Google ScholarLocate open access versionFindings
  • S. Bach, A. Binder, G. Montavon, F. Klauschen, K.-R. Muller, and W. Samek. On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation. PloS one, 10(7):e0130140, 2015.
    Google ScholarLocate open access versionFindings
  • D. Bahdanau, K. Cho, and Y. Bengio. Neural machine translation by jointly learning to align and translate. arXiv preprint arXiv:1409.0473, 2014.
    Findings
  • O. Biran and C. Cotton. Explanation and justification in machine learning: A survey. In IJCAI-17 Workshop on Explainable AI (XAI), p. 8, 2017.
    Google ScholarLocate open access versionFindings
  • D. Cashman, G. Patterson, A. Mosca, and R. Chang. Rnnbow: Visualizing learning via backpropagation gradients in recurrent neural networks. In Workshop on Visual Analytics for Deep Learning (VADL), 2017.
    Google ScholarFindings
  • Y. Ding, Y. Liu, H. Luan, and M. Sun. Visualizing and understanding neural machine translation. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), vol. 1, pp. 1150–1159, 2017.
    Google ScholarLocate open access versionFindings
  • D. Erhan, Y. Bengio, A. Courville, and P. Vincent. Visualizing higherlayer features of a deep network. Technical report, University of Montreal, 2009.
    Google ScholarFindings
  • J. Gehring, M. Auli, D. Grangier, and Y. N. Dauphin. A Convolutional Encoder Model for Neural Machine Translation. ArXiv e-prints, Nov. 2016.
    Google ScholarFindings
  • J. Gehring, M. Auli, D. Grangier, D. Yarats, and Y. N. Dauphin. Convolutional Sequence to Sequence Learning. ArXiv e-prints, May 2017.
    Google ScholarFindings
  • H. Hassan Awadalla, A. Aue, C. Chen, V. Chowdhary, J. Clark, C. Federmann, X. Huang, M. Junczys-Dowmunt, W. Lewis, M. Li, S. Liu, T.-Y. Liu, R. Luo, A. Menezes, T. Qin, F. Seide, X. Tan, F. Tian, L. Wu, S. Wu, Y. Xia, D. Zhang, Z. Zhang, and M. Zhou. Achieving human parity on automatic chinese to english news translation. March 2018.
    Google ScholarFindings
  • R. Henderson and R. Rothe. Picasso: A modular framework for visualizing the learning process of neural network image classifiers. Journal of Open Research Software, 5(1), 2017.
    Google ScholarLocate open access versionFindings
  • A. Hern. Facebook translates ’good morning’ into ’attack them’, leading to arrest. The Guardian, Oct 2017.
    Google ScholarLocate open access versionFindings
  • F. Hohman, M. Kahng, R. Pienta, and D. H. Chau. Visual analytics in deep learning: An interrogative survey for the next frontiers. arXiv preprint arXiv:1801.06889, 2018.
    Findings
  • J. Johnson, M. Douze, and H. Jegou. Billion-scale similarity search with gpus. arXiv preprint arXiv:1702.08734, 2017.
    Findings
  • M. Johnson, M. Schuster, Q. V. Le, M. Krikun, Y. Wu, Z. Chen, N. Thorat, F. Viegas, M. Wattenberg, G. Corrado, et al. Google’s multilingual neural machine translation system: enabling zero-shot translation. arXiv preprint arXiv:1611.04558, 2016.
    Findings
  • M. Kahng, P. Y. Andrews, A. Kalro, and D. H. P. Chau. Activis: Visual exploration of industry-scale deep neural network models. IEEE transactions on visualization and computer graphics, 24(1):88–97, 2018.
    Google ScholarLocate open access versionFindings
  • A. Karpathy, J. Johnson, and F.-F. Li. Visualizing and understanding recurrent networks. ICLR Workshops, 2015.
    Google ScholarLocate open access versionFindings
  • G. Klein, Y. Kim, Y. Deng, J. Senellart, and A. M. Rush. OpenNMT: Open-Source Toolkit for Neural Machine Translation. ArXiv e-prints.
    Google ScholarFindings
  • P. Koehn and R. Knowles. Six challenges for neural machine translation. arXiv preprint arXiv:1706.03872, 2017.
    Findings
  • P. W. Koh and P. Liang. Understanding black-box predictions via influence functions. arXiv preprint arXiv:1703.04730, 2017.
    Findings
  • J. Krause, A. Perer, and K. Ng. Interacting with predictions: Visual inspection of black-box machine learning models. In Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems, pp. 5686–5697. ACM, 2016.
    Google ScholarLocate open access versionFindings
  • J. B. Kruskal. Nonmetric multidimensional scaling: a numerical method. Psychometrika, 29(2):115–129, 1964.
    Google ScholarLocate open access versionFindings
  • Q. V. Le, M. Ranzato, R. Monga, M. Devin, G. Corrado, K. C. 0010, J. Dean, and A. Y. Ng. Building high-level features using large scale unsupervised learning. ICML, 2012.
    Google ScholarLocate open access versionFindings
  • Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner. Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11):2278– 2324, 1998.
    Google ScholarLocate open access versionFindings
  • J. Lee, J.-H. Shin, and J.-S. Kim. Interactive visualization and manipulation of attention-based neural machine translation. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pp. 121–126, 2017.
    Google ScholarLocate open access versionFindings
  • T. Lei, R. Barzilay, and T. S. Jaakkola. Rationalizing neural predictions. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, EMNLP 2016, Austin, Texas, USA, November 1-4, 2016, pp. 107–117, 2016.
    Google ScholarLocate open access versionFindings
  • J. Li, X. Chen, E. Hovy, and D. Jurafsky. Visualizing and Understanding Neural Models in NLP. In NAACL, pp. 1–10. Association for Computational Linguistics, San Diego, California, jun 2016.
    Google ScholarLocate open access versionFindings
  • J. Li, W. Monroe, and D. Jurafsky. Understanding neural networks through representation erasure. arXiv preprint arXiv:1612.08220, 2016.
    Findings
  • P. J. Liu, M. Saleh, E. Pot, B. Goodrich, R. Sepassi, L. Kaiser, and N. Shazeer. Generating wikipedia by summarizing long sequences. arXiv preprint arXiv:1801.10198, 2018.
    Findings
  • L. v. d. Maaten and G. Hinton. Visualizing data using t-sne. Journal of machine learning research, 9(Nov):2579–2605, 2008.
    Google ScholarLocate open access versionFindings
  • C. Mauro, G. Christian, and F. Marcello. Wit3: Web inventory of transcribed and translated talks. In Conference of European Association for Machine Translation, pp. 261–268, 2012.
    Google ScholarLocate open access versionFindings
  • Y. Ming, S. Cao, R. Zhang, Z. Li, Y. Chen, Y. Song, and H. Qu. Understanding hidden memories of recurrent neural networks. arXiv preprint arXiv:1710.10777, 2017.
    Findings
  • R. Nallapati, B. Zhou, C. Gulcehre, B. Xiang, et al. Abstractive text summarization using sequence-to-sequence rnns and beyond. arXiv preprint arXiv:1602.06023, 2016.
    Findings
  • M. Narayanan, E. Chen, J. He, B. Kim, S. Gershman, and F. Doshi-Velez. How do humans understand explanations from machine learning systems? an evaluation of the human-interpretability of explanation. arXiv preprint arXiv:1802.00682, 2018.
    Findings
  • A. Nguyen, J. Yosinski, and J. Clune. Deep neural networks are easily fooled: High confidence predictions for unrecognizable images. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 427–436, 2015.
    Google ScholarLocate open access versionFindings
  • C. Olah and S. Carter. Attention and augmented recurrent neural networks. Distill, 2016. doi: 10.23915/distill.00001
    Locate open access versionFindings
  • C. Olah, A. Satyanarayan, I. Johnson, S. Carter, L. Schubert, K. Ye, and A. Mordvintsev. The building blocks of interpretability. Distill, 3(3):e10, 2018.
    Google ScholarLocate open access versionFindings
  • D. Park, S. Kim, J. Lee, J. Choo, N. Diakopoulos, and N. Elmqvist. Conceptvector: text visual analytics via interactive lexicon building using word embedding. IEEE transactions on visualization and computer graphics, 24(1):361–370, 2018.
    Google ScholarLocate open access versionFindings
  • R. Paulus, C. Xiong, and R. Socher. A deep reinforced model for abstractive summarization. arXiv preprint arXiv:1705.04304, 2017.
    Findings
  • F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay. Scikitlearn: Machine learning in Python. Journal of Machine Learning Research, 12:2825–2830, 2011.
    Google ScholarLocate open access versionFindings
  • M. T. Ribeiro, S. Singh, and C. Guestrin. Why should i trust you?: Explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1135–1144. ACM, 2016.
    Google ScholarLocate open access versionFindings
  • A. S. Ross, M. C. Hughes, and F. Doshi-Velez. Right for the right reasons: Training differentiable models by constraining their explanations. arXiv preprint arXiv:1703.03717, 2017.
    Findings
  • A. Ruckleand I. Gurevych. End-to-end non-factoid question answering with an interactive visualization of neural attention weights. Proceedings of ACL 2017, System Demonstrations, pp. 19–24, 2017.
    Google ScholarLocate open access versionFindings
  • A. M. Rush, S. Chopra, and J. Weston. A neural attention model for abstractive sentence summarization. arXiv preprint arXiv:1509.00685, 2015.
    Findings
  • A. See, P. J. Liu, and C. D. Manning. Get to the point: Summarization with pointer-generator networks. arXiv preprint arXiv:1704.04368, 2017.
    Findings
  • K. Simonyan, A. Vedaldi, and A. Zisserman. Deep inside convolutional networks: Visualising image classification models and saliency maps. arXiv preprint arXiv:1312.6034, 2013.
    Findings
  • D. Smilkov, S. Carter, D. Sculley, F. B. Viegas, and M. Wattenberg. Direct-manipulation visualization of deep networks. arXiv preprint arXiv:1708.03788, 2017.
    Findings
  • H. Strobelt, S. Gehrmann, H. Pfister, and A. M. Rush. Lstmvis: A tool for visual analysis of hidden state dynamics in recurrent neural networks. IEEE transactions on visualization and computer graphics, 24(1):667–676, 2018.
    Google ScholarLocate open access versionFindings
  • I. Sutskever, O. Vinyals, and Q. V. Le. Sequence to sequence learning with neural networks. In Advances in neural information processing systems, pp. 3104–3112, 2014.
    Google ScholarLocate open access versionFindings
  • A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin. Attention is all you need. In Advances in Neural Information Processing Systems, pp. 6000–6010, 2017.
    Google ScholarLocate open access versionFindings
  • Y. Wu, M. Schuster, Z. Chen, Q. V. Le, M. Norouzi, W. Macherey, M. Krikun, Y. Cao, Q. Gao, K. Macherey, et al. Google’s neural machine translation system: Bridging the gap between human and machine translation. arXiv preprint arXiv:1609.08144, 2016.
    Findings
  • K. Xu, J. Ba, R. Kiros, K. Cho, A. Courville, R. Salakhudinov, R. Zemel, and Y. Bengio. Show, attend and tell: Neural image caption generation with visual attention. In International Conference on Machine Learning, pp. 2048–2057, 2015.
    Google ScholarLocate open access versionFindings
  • J. Yosinski, J. Clune, A. Nguyen, T. Fuchs, and H. Lipson. Understanding neural networks through deep visualization. arXiv preprint arXiv:1506.06579, 2015.
    Findings
  • M. D. Zeiler and R. Fergus. Visualizing and Understanding Convolutional Networks. In Computer Vision–ECCV, vol. 8689, pp. 818–833. Springer, 2014. doi: 10.1007/978-3-319-10590-1 53
    Locate open access versionFindings
  • L. M. Zintgraf, T. S. Cohen, T. Adel, and M. Welling. Visualizing deep neural network decisions: Prediction difference analysis. ICML, 2017.
    Google ScholarLocate open access versionFindings
0
Your rating :

No Ratings

Tags
Comments
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn