Abstract
In this demonstration, we present Krypton, a system for accelerating occlusion-based deep convolution neural network (CNN) explanation workloads. Driven by the success of CNNs in image understanding tasks, there is growing adoption of CNNs in various domains, including high stakes applications such as radiology. However, users of such applications often seek an "explanation" for why a CNN predicted a certain label. One of the most widely used approaches for explaining CNN predictions is the occlusion-based explanation (OBE) method. This approach is computationally expensive due to the large number of re-inference requests produced. Krypton reduces the runtime of OBE by up to 35x by enabling incremental and approximate inference optimizations that are inspired by classical database query optimization techniques. We allow the audience to interactively diagnose CNN predictions from several use cases, including radiology and natural images. A short video of our demonstration can be found here: https://youtu.be/1OWddbd4n6Y
- Ai device for detecting diabetic retinopathy earns swift fda approval. https://www.aao.org/headline/first-ai-screen-diabetic-retinopathy-approved-by-f. Accessed March 15, 2019.Google Scholar
- Incremental and approximate inference for faster occlusion-based deep cnn explanations. https://adalabucsd.github.io/papers/TR_2019_Krypton.pdf. Accessed March 15, 2019.Google Scholar
- M. Buckler et al. Eva<sup>2</sup>: Exploiting temporal redundancy in live computer vision. arXiv preprint arXiv:1803.06312, 2018. Google ScholarDigital Library
- L. Cavigelli et al. Cbinfer: Change-based inference for convolutional neural networks on video data. In 11th International Conference on Distributed Smart Cameras, pages 1--8. ACM, 2017. Google ScholarDigital Library
- R. Chirkova, J. Yang, et al. Materialized views. Foundations and Trends in Databases, 4(4):295--405, 2012. Google ScholarCross Ref
- K.-H. Jung et al. Deep learning for medical image analysis: Applications to computed tomography and magnetic resonance imaging. Hanyang Medical Reviews, 37(2):61--70, 2017.Google ScholarCross Ref
- D. S. Kermany et al. Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell, 172(5):1122--1131, 2018.Google ScholarCross Ref
- T. Miller. Explanation in artificial intelligence: Insights from the social sciences. arXiv preprint arXiv:1706.07269, 2017.Google Scholar
- M. T. Ribeiro et al. Why should i trust you?: Explaining the predictions of any classifier. In 22nd ACM SIGKDD, pages 1135--1144. ACM, 2016. Google ScholarDigital Library
- T. K. Sellis. Multiple-query optimization. ACM TODS, 13(1):23--52, 1988. Google ScholarDigital Library
- K. Simonyan et al. Deep inside convolutional networks: Visualising image classification models and saliency maps. arXiv preprint arXiv:1312.6034, 2013.Google Scholar
- P. Voigt and A. Von dem Bussche. The EU General Data Protection Regulation, volume 18. Springer, 2017. Google ScholarDigital Library
- M. D. Zeiler and R. Fergus. Visualizing and understanding convolutional networks. In ECCV, pages 818--833. Springer, 2014.Google ScholarCross Ref
- L. M. Zintgraf et al. Visualizing deep neural network decisions: Prediction difference analysis. arXiv preprint arXiv:1702.04595, 2017.Google Scholar
Recommendations
Live Demonstration: Deep Learning-Based Multiple Object Detection and Tracking on a Low-Power Embedded System
ICDSC '18: Proceedings of the 12th International Conference on Distributed Smart CamerasDespite the hopeful prospect for image processing using Convolutional Neural Netwoks, CNNs, the gap between software and hardware solutions is already considerable for embedded applications due to their high power consumption. This demo performs low-...
Undecimated wavelet shrinkage estimate of the 1D and 2D spectra
ICASSP '00: Proceedings of the Acoustics, Speech, and Signal Processing, 2000. on IEEE International Conference - Volume 04We study the problem of estimating the log-spectrum of a stationary Gaussian time series by thresholding the wavelet coefficients. We propose the use of the undecimated wavelet transform to denoise the log-periodogram. For this, we review a denoising ...
A Demonstration of GeoTorchAI: A Spatiotemporal Deep Learning Framework
SIGMOD '23: Companion of the 2023 International Conference on Management of DataThis paper demonstrates GeoTorchAI, a spatiotemporal deep learning framework. In recent years, many neural network models have been proposed focusing on the applications of raster imagery and spatiotemporal non-imagery datasets. Implementing these ...
Comments