检索结果-内蒙古大学图书馆

Better Understanding Differences in Attribution Methods via Systematic Evaluations

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Rao, Sukrut Böhle, Moritz Schiele, Bernt The Department of Computer Vision and Machine Learning Max Planck Institute for Informatics Saarland Informatics Campus Saarbrücken66123 Germany

Deep neural networks are very successful on many vision tasks, but hard to interpret due to their black box nature. To overcome this, various post-hoc attribution methods have been proposed to identify image regions most influential to the models' decisions. Evaluating such methods is challenging since no ground truth attributions exist. We thus propose three novel evaluation schemes to more reliably measure the faithfulness of those methods, to make comparisons between them more fair, and to make visual inspection more systematic. To address faithfulness, we propose a novel evaluation setting (DiFull) in which we carefully control which parts of the input can influence the output in order to distinguish possible from impossible attributions. To address fairness, we note that different methods are applied at different layers, which skews any comparison, and so evaluate all methods on the same layers (ML-Att) and discuss how this impacts their performance on quantitative metrics. For more systematic visualizations, we propose a scheme (AggAtt) to qualitatively evaluate the methods on complete datasets. We use these evaluation schemes to study strengths and shortcomings of some widely used attribution methods over a wide range of models. Finally, we propose a post-processing smoothing step that significantly improves the performance of some attribution methods, and discuss its applicability. Copyright © 2023, The Authors. All rights reserved.

关键词： Deep neural networks

Overcoming multi-model forgetting

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Benyahia, Yassine Yu, Kaicheng Bennani-Smires, Kamil Jaggi, Martin Davison, Anthony Salzmann, Mathieu Musat, Claudiu Institute of mathematics EPFL Computer vision lab EPFL Artificial Intellegence Lab Swisscom Machine learning and optimization lab EPFL

We identify a phenomenon, which we refer to as multi-model forgetting, that occurs when sequentially training multiple deep networks with partially-shared parameters;the performance of previously-trained models degrades as one optimizes a subsequent one, due to the overwriting of shared parameters. To overcome this, we introduce a statistically-justified weight plasticity loss that regularizes the learning of a model's shared parameters according to their importance for the previous models, and demonstrate its effectiveness when training two models sequentially and for neural architecture search. Adding weight plasticity in neural architecture search preserves the best models to the end of the search and yields improved results in both natural language processing and computer vision tasks. Copyright © 2019, The Authors. All rights reserved.

关键词： Natural language processing systems

ActMAD: Activation Matching to Align Distributions for Test-Time-Training

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Mirza, Muhammad Jehanzeb Soneira, Pol Jané Lin, Wei Kozinski, Mateusz Possegger, Horst Bischof, Horst Institute for Computer Graphics and Vision TU Graz Austria Christian Doppler Laboratory for Embedded Machine Learning Institute of Control Systems KIT Germany Christian Doppler Laboratory for Semantic 3D Computer Vision

Test-Time-Training (TTT) is an approach to cope with out-of-distribution (OOD) data by adapting a trained model to distribution shifts occurring at test-time. We propose to perform this adaptation via Activation Matching (ActMAD): We analyze activations of the model and align activation statistics of the OOD test data to those of the training data. In contrast to existing methods, which model the distribution of entire channels in the ultimate layer of the feature extractor, we model the distribution of each feature in multiple layers across the network. This results in a more fine-grained supervision and makes ActMAD attain state of the art performance on CIFAR-100C and Imagenet-C. ActMAD is also architecture- and task-agnostic, which lets us go beyond image classification, and score 15.4% improvement over previous approaches when evaluating a KITTI-trained object detector on KITTI-Fog. Our experiments highlight that ActMAD can be applied to online adaptation in realistic scenarios, requiring little data to attain its full performance. © 2022, CC BY.

关键词： Chemical activation

ActMAD: Activation Matching to Align Distributions for Test-Time-Training

学校读者我要写书评

暂无评论

ActMAD: Activation Matching to Align Distributions for Test-...

Conference on computer vision and Pattern Recognition (CVPR)

作者： M. Jehanzeb Mirza Pol Jané Soneira Wei Lin Mateusz Kozinski Horst Possegger Horst Bischof Institute for Computer Graphics and Vision TU Graz Austria Christian Doppler Laboratory for Embedded Machine Learning Institute of Control Systems KIT Germany Christian Doppler Laboratory for Semantic 3D Computer Vision

Test-Time-Training (TTT) is an approach to cope with out-of-distribution (OOD) data by adapting a trained model to distribution shifts occurring at test-time. We propose to perform this adaptation via Activation Matching (ActMAD): We analyze activations of the model and align activation statistics of the OOD test data to those of the training data. In contrast to existing methods, which model the distribution of entire channels in the ultimate layer of the feature extractor, we model the distribution of each feature in multiple layers across the network. This results in a more fine-grained supervision and makes ActMAD attain state of the art performance on CIFAR-100C and Imagenet-C. ActMAD is also architecture-and task-agnostic, which lets us go beyond image classification, and score 15.4% improvement over previous approaches when evaluating a KITTI-trained object detector on KITTI-Fog. Our experiments highlight that ActMAD can be applied to online adaptation in realistic scenarios, requiring little data to attain its full performance.

关键词：

Pruning neural network models for gene regulatory dynamics using data and domain knowledge 38

学校读者我要写书评

暂无评论

Pruning neural network models for gene regulatory dynamics u...

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Hossain, Intekhab Fischer, Jonas Burkholz, Rebekka Quackenbush, John Department of Biostatistics Harvard T.H. Chan School of Public Health BostonMA02115 United States Dep. for Computer Vision and Machine Learning Max Planck Institute for Informatics Saarbrücken Germany Helmholtz Center CISPA for Information Security Saarbrücken Germany

The practical utility of machine learning models in the sciences often hinges on their interpretability. It is common to assess a model's merit for scientific discovery, and thus novel insights, by how well it aligns with already available domain knowledge-a dimension that is currently largely disregarded in the comparison of neural network models. While pruning can simplify deep neural network architectures and excels in identifying sparse models, as we show in the context of gene regulatory network inference, state-of-the-art techniques struggle with biologically meaningful structure learning. To address this issue, we propose DASH‡, a generalizable framework that guides network pruning by using domain-specific structural information in model fitting and leads to sparser, better interpretable models that are more robust to noise. Using both synthetic data with ground truth information, as well as real-world gene expression data, we show that DASH, using knowledge about gene interaction partners within the putative regulatory network, outperforms general pruning methods by a large margin and yields deeper insights into the biological systems being studied. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

Robustness of object detectors in degrading weather conditions

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Mirza, Muhammad Jehanzeb Buerkle, Cornelius Jarquin, Julio Opitz, Michael Oboril, Fabian Scholl, Kay-Ulrich Bischof, Horst Christian Doppler Laboratory for Embedded Machine Learning Institute for Computer Graphics and Vision Graz University of Technology Austria Intel Labs Karlsruhe Germany

State-of-the-art object detection systems for autonomous driving achieve promising results in clear weather conditions. However, such autonomous safety critical systems also need to work in degrading weather conditions, such as rain, fog and snow. Unfortunately, most approaches evaluate only on the KITTI dataset, which consists only of clear weather scenes. In this paper we address this issue and perform one of the most detailed evaluation on single and dual modality architectures on data captured in real weather conditions. We analyse the performance degradation of these architectures in degrading weather conditions. We demonstrate that an object detection architecture performing good in clear weather might not be able to handle degrading weather conditions. We also perform ablation studies on the dual modality architectures and show their limitations. © 2021, CC BY.

关键词： Object detection

MATE: Masked Autoencoders are Online 3D Test-Time Learners

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Mirza, M. Jehanzeb Shin, Inkyu Lin, Wei Schriebl, Andreas Sun, Kunyang Choe, Jaesung Possegger, Horst Kozinski, Mateusz Kweon, In So Yoon, Kuk-Jin Bischof, Horst Institute for Computer Graphics and Vision Graz University of Technology Austria Christian Doppler Laboratory for Embedded Machine Learning Korea Republic of Southeast University China

Our MATE is the first Test-Time-Training (TTT) method designed for 3D data, which makes deep networks trained for point cloud classification robust to distribution shifts occurring in test data. Like existing TTT methods from the 2D image domain, MATE also leverages test data for adaptation. Its test-time objective is that of a Masked Autoencoder: a large portion of each test point cloud is removed before it is fed to the network, tasked with reconstructing the full point cloud. Once the network is updated, it is used to classify the point cloud. We test MATE on several 3D object classification datasets and show that it significantly improves robustness of deep networks to several types of corruptions commonly occurring in 3D point clouds. We show that MATE is very efficient in terms of the fraction of points it needs for the adaptation. It can effectively adapt given as few as 5% of tokens of each test sample, making it extremely lightweight. Our experiments show that MATE also achieves competitive performance by adapting sparsely on the test data, which further reduces its computational overhead, making it ideal for real-time applications. © 2022, CC BY.

关键词： Classification (of information)

Brain tumor cell density estimation from multi-modal MR images based on a synthetic tumor growth model

学校读者我要写书评

暂无评论

Brain tumor cell density estimation from multi-modal MR imag...

15th International Conference on Medical Image Computing and computer-Assisted Intervention, MICCAI 2012

作者： Geremia, Ezequiel Menze, Bjoern H. Prastawa, Marcel Weber, M.-A. Criminisi, Antonio Ayache, Nicholas Asclepios Research Project INRIA Sophia-Antipolis France Computer Science and Artificial Intelligence Laboratory MIT United States Computer Vision Laboratory ETH Zurich Switzerland Scientific Computing and Imaging Institute University of Utah United States Diagnostic and Interventional Radiology Heidelberg University Hospital Germany Machine Learning and Perception Group Microsoft Research Cambridge United Kingdom

ISBN: (纸本)9783642366192

This paper proposes to employ a detailed tumor growth model to synthesize labelled images which can then be used to train an efficient data-driven machine learning tumor predictor. Our MR image synthesis step generates images with both healthy tissues as well as various tumoral tissue types. Subsequently, a discriminative algorithm based on random regression forests is trained on the simulated ground truth to predict the continuous latent tumor cell density, and the discrete tissue class associated with each voxel. The presented method makes use of a large synthetic dataset of 740 simulated cases for training and evaluation. A quantitative evaluation on 14 real clinical cases diagnosed with low-grade gliomas demonstrates tissue class accuracy comparable with state of the art, with added benefit in terms of computational efficiency and the ability to estimate tumor cell density as a latent variable underlying the multimodal image observations. The idea of synthesizing training data to train data-driven learning algorithms can be extended to other applications where expert annotation is lacking or expensive. © 2013 Springer-Verlag.

关键词： Tumors

DWDN: Deep Wiener Deconvolution Network for Non-Blind Image Deblurring

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Dong, Jiangxin Roth, Stefan Schiele, Bernt School of Computer Science and Engineering Nanjing University of Science and Technology China Department of Computer Vision and Machine Learning Max Planck Institute for Informatics Germany Department of Computer Science TU Darmstadt Germany

We present a simple and effective approach for non-blind image deblurring, combining classical techniques and deep learning. In contrast to existing methods that deblur the image directly in the standard image space, we propose to perform an explicit deconvolution process in a feature space by integrating a classical Wiener deconvolution framework with learned deep features. A multi-scale cascaded feature refinement module then predicts the deblurred image from the deconvolved deep features, progressively recovering detail and small-scale structures. The proposed model is trained in an end-to-end manner and evaluated on scenarios with simulated Gaussian noise, saturated pixels, or JPEG compression artifacts as well as real-world images. Moreover, we present detailed analyses of the benefit of the feature-based Wiener deconvolution and of the multi-scale cascaded feature refinement as well as the robustness of the proposed approach. Our extensive experimental results show that the proposed deep Wiener deconvolution network facilitates deblurred results with visibly fewer artifacts and quantitatively outperforms state-of-the-art non-blind image deblurring methods by a wide margin. Copyright © 2021, The Authors. All rights reserved.

关键词： Deconvolution