Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
184 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

ProtoExplorer: Interpretable Forensic Analysis of Deepfake Videos using Prototype Exploration and Refinement (2309.11155v1)

Published 20 Sep 2023 in cs.AI

Abstract: In high-stakes settings, Machine Learning models that can provide predictions that are interpretable for humans are crucial. This is even more true with the advent of complex deep learning based models with a huge number of tunable parameters. Recently, prototype-based methods have emerged as a promising approach to make deep learning interpretable. We particularly focus on the analysis of deepfake videos in a forensics context. Although prototype-based methods have been introduced for the detection of deepfake videos, their use in real-world scenarios still presents major challenges, in that prototypes tend to be overly similar and interpretability varies between prototypes. This paper proposes a Visual Analytics process model for prototype learning, and, based on this, presents ProtoExplorer, a Visual Analytics system for the exploration and refinement of prototype-based deepfake detection models. ProtoExplorer offers tools for visualizing and temporally filtering prototype-based predictions when working with video data. It disentangles the complexity of working with spatio-temporal prototypes, facilitating their visualization. It further enables the refinement of models by interactively deleting and replacing prototypes with the aim to achieve more interpretable and less biased predictions while preserving detection accuracy. The system was designed with forensic experts and evaluated in a number of rounds based on both open-ended think aloud evaluation and interviews. These sessions have confirmed the strength of our prototype based exploration of deepfake videos while they provided the feedback needed to continuously improve the system.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (63)
  1. Deepfacelab: Integrated, flexible and extensible face-swapping framework. arXiv preprint arXiv:200505535 2020; .
  2. Reface — Face swap videos. https://hey.reface.ai/.
  3. Avatarify — Bring your photos to life. https://avatarify.ai/.
  4. Dolhansky B, Bitton J and Pflaum ea Ben. The deepfake detection challenge (dfdc) dataset. arXiv preprint arXiv:200607397 2020; .
  5. Westerlund M. The emergence of deepfake technology: A review. Technol Innov Manag Rev 2019; 9.
  6. Rossler A, Cozzolino D and Verdoliva Lea. Faceforensics++: Learning to detect manipulated facial images. In Proceedings of the IEEE/CVF international conference on computer vision. pp. 1–11.
  7. Verdoliva L. Media Forensics and DeepFakes: An Overview. IEEE J Sel Top Signal Process 2020; 14(5): 910–932.
  8. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization. In Proceedings of the IEEE International Conference on Computer Vision. pp. 618–626.
  9. Layer-wise relevance propagation for neural networks with local renormalization layers. In Proceedings of Artificial Neural Networks and Machine Learning (ICANN). Springer, pp. 63–71.
  10. Shrikumar A, Greenside P and Kundaje A. Learning Important Features Through Propagating Activation Differences. In Precup D and Teh YW (eds.) Proceedings of the 34th International Conference on Machine Learning, Proceedings of Machine Learning Research, volume 70. PMLR, pp. 3145–3153.
  11. Sundararajan M, Taly A and Yan Q. Axiomatic Attribution for Deep Networks. In Precup D and Teh YW (eds.) Proceedings of the 34th International Conference on Machine Learning, Proceedings of Machine Learning Research, volume 70. PMLR, pp. 3319–3328.
  12. Ribeiro MT, Singh S and Guestrin C. 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.
  13. Explainable deep learning: A field guide for the uninitiated. J Artif Intell Res 2022; 73: 329–396.
  14. Case-Based reasoning: Foundational issues, methodological variations, and system approaches. AI Commun 1994; 7(1).
  15. Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges. Stat Surv 2021; .
  16. Human Problem Solving . Englewood Cliffs, NJ: Prentice-Hall, 1972.
  17. Edelman S. Representation of Similarity in Three-Dimensional Object Discrimination. Neural Comput 1995; 7(2): 408–423.
  18. Orientation dependence in the recognition of familiar and novel views of three-dimensional objects. Vision Res 1992; 32: 2385–2400.
  19. Ware C. Information Visualization. Perception for Design. Morgan Kaufmann Publishers, 2004.
  20. Interpretable and trustworthy deepfake detection via dynamic prototypes. In Proceedings of the IEEE/CVF winter conference on applications of computer vision. pp. 1973–1983.
  21. Deep learning for case-based reasoning through prototypes: A neural network that explains its predictions. In Proceedings of the AAAI Conference on Artificial Intelligence. pp. 3530–3537.
  22. This Looks Like That: Deep Learning for Interpretable Image Recognition. Adv Neural Inf Process Syst 2019; 32.
  23. Migut MA, van Gemert JC and Worring M. Interactive decision making using dissimilarity to visually represented prototypes. In 2011 IEEE Conference on Visual Analytics Science and Technology (VAST). IEEE, pp. 141–149.
  24. ProtoViewer: Visual Interpretation and Diagnostics of Deep Neural Networks with Factorized Prototypes. In 2020 IEEE Visualization Conference (VIS). Salt Lake City, UT, USA: IEEE. ISBN 978-1-7281-8015-1, pp. 286–290.
  25. Interpreting Deep Neural Networks through Prototype Factorization. In 2020 International Conference on Data Mining Workshops (ICDMW). pp. 448–457.
  26. A survey of visual analytics for explainable artificial intelligence methods. Comput Graph 2022; 102: 502–520.
  27. A survey of visual analytics techniques for machine learning. Comput Vis Media 2021; 7(1).
  28. Protosteer: Steering deep sequence model with prototypes. IEEE Trans Vis Comput Graph 2020; 26(1): 238–248.
  29. Visual auditor: Interactive visualization for detection and summarization of model biases. In 2022 IEEE Visualization and Visual Analytics (VIS). pp. 45–49.
  30. RMExplorer: A Visual Analytics Approach to Explore the Performance and the Fairness of Disease Risk Models on Population Subgroups. In 2022 IEEE Visualization and Visual Analytics (VIS). pp. 50–54.
  31. An Examination of Fairness of AI Models for Deepfake Detection. arXiv preprint arXiv:210500558 2021; .
  32. Molnar C. Interpretable Machine Learning. 2 ed. 2022.
  33. Miller T. Explanation in artificial intelligence: Insights from the social sciences. Artif Intell 2019; 267: 1–38.
  34. Sanity checks for saliency maps. Adv Neural Inf Process Syst 2018; 31.
  35. Nauta M, van Bree R and Seifert C. Neural Prototype Trees for Interpretable Fine-Grained Image Recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, pp. 14933–14943.
  36. Interpretable Image Recognition with Hierarchical Prototypes. In Proceedings of the AAAI Conference on Human Computation and Crowdsourcing, volume 7. pp. 32–40.
  37. Donnelly J, Barnett AJ and Chen C. Deformable ProtoPNet: An Interpretable Image Classifier Using Deformable Prototypes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 10265–10275.
  38. XProtoNet: Diagnosis in Chest Radiography With Global and Local Explanations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 15719–15728.
  39. Interpretable and Steerable Sequence Learning via Prototypes. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. pp. 903–913.
  40. explAIner: A Visual Analytics Framework for Interactive and Explainable Machine Learning. IEEE Trans Vis Comput Graph 2019; .
  41. Collaris D and van Wijk JJ. ExplainExplore: Visual Exploration of Machine Learning Explanations. In Beck F, Seo J and Wang C (eds.) 2020 IEEE Pacific Visualization Symposium, PacificVis 2020 - Proceedings. United States: Institute of Electrical and Electronics Engineers, pp. 26–35.
  42. Human-in-the-loop Extraction of Interpretable Concepts in Deep Learning Models. IEEE Trans Vis Comput Graph 2022; 28(1): 780–790.
  43. Alperin KB, Wollaber AB and Gomez SR. Improving Interpretability for Cyber Vulnerability Assessment Using Focus and Context Visualizations. In 2020 IEEE Symposium on Visualization for Cyber Security (VizSec). IEEE, pp. 30–39.
  44. Wang J, Zhang W and Yang H. SCANViz: Interpreting the Symbol-Concept Association Captured by Deep Neural Networks through Visual Analytics. In 2020 IEEE Pacific Visualization Symposium (PacificVis). Los Alamitos, CA, USA: IEEE Computer Society, pp. 51–60.
  45. Summit: Scaling Deep Learning Interpretability by Visualizing Activation and Attribution Summarizations. IEEE Trans Vis Comput Graph 2020; 26(01): 1096–1106.
  46. VATLD: A Visual Analytics System to Assess, Understand and Improve Traffic Light Detection. IEEE Trans Vis Comput Graph 2021; 27(2): 261–271.
  47. VBridge: Connecting the Dots Between Features and Data to Explain Healthcare Models. IEEE Trans Vis Comput Graph 2021; : 1–1.
  48. A visual analytics system for improving attention-based traffic forecasting models. IEEE Trans Vis Comput Graph 2022; 29(1): 1102–1112.
  49. Deep High-Resolution Representation Learning for Visual Recognition. IEEE Trans Pattern Anal Mach Intell 2021; 43(10): 3349–3364.
  50. Knowledge Generation Model for Visual Analytics. IEEE Trans Vis Comput Graph 2014; 20(12): 1604–1613.
  51. Visual analytics: Definition, process, and challenges. Springer, 2008.
  52. Ware C. Visual Thinking for Design. Morgan Kaufmann Publishers, 2008.
  53. McInnes L, Healy J and Melville J. UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction. arXiv preprint arXiv:180203426 2018; .
  54. This looks more like that: Enhancing self-explaining models by prototypical relevance propagation. Pattern Recognit 2023; 136: 109172.
  55. On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation. PloS one 2015; 10(7): e0130140.
  56. DeepFlow: Large Displacement Optical Flow with Deep Matching. In Proceedings of the IEEE international conference on computer vision. pp. 1385–1392.
  57. ImageNet: A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition. pp. 248–255.
  58. Deepfakes GitHub. https://github.com/deepfakes/faceswap.
  59. Thies J, Zollhöfer M and Nießner M. Deferred neural rendering: image synthesis using neural textures. ACM Trans Graph 2019; 38(4): 66:1–66:12.
  60. Faceswap GitHub. https://github.com/MarekKowalski/FaceSwap/.
  61. Face2Face: Real-time Face Capture and Reenactment of RGB Videos. In Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 2387–2395.
  62. Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks. IEEE Signal Process Lett 2016; 23(10): 1499–1503.
  63. Jakob Nielsen. Usability Engineering. Cambridge MA: Academic Press, 1993.
Citations (7)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now