Convolutional Dynamic Alignment Networks for Interpretable Classifications

2021-03-31 · CVPR 2021 1 · Moritz Böhle, Mario Fritz, Bernt Schiele

We introduce a new family of neural network models called Convolutional Dynamic Alignment Networks (CoDA-Nets), which are performant classifiers with a high degree of inherent interpretability. Their core building blocks are Dynamic Alignment Units (DAUs), which linearly transform their input with weight vectors that dynamically align with task-relevant patterns. As a result, CoDA-Nets model the classification prediction through a series of input-dependent linear transformations, allowing for linear decomposition of the output into individual input contributions. Given the alignment of the DAUs, the resulting contribution maps align with discriminative input patterns. These model-inherent decompositions are of high visual quality and outperform existing attribution methods under quantitative metrics. Further, CoDA-Nets constitute performant classifiers, achieving on par results to ResNet and VGG models on e.g. CIFAR-10 and TinyImagenet.

📄 PDF Abstract BibTeX arXiv:2104.00032

Code (1)

moboehle/CoDA-Nets 공식 구현 pytorch

Methods 이 논문이 사용한 방법론

1x1 Convolution A 1 x 1 Convolution is a convolution with some special properties in that it can be used for dimensionality reduction,…

Average Pooling 설명 없음

Batch Normalization 설명 없음

Global Average Pooling Global Average Pooling is a pooling operation designed to replace fully connected layers in classical CNNs. The idea is to generate one feature map for each corresponding…

Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…

Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…

Dropout Dropout is a regularization technique for neural networks that drops a unit (along with connections) at training time with a specified probability $p$ (a common value is…

Max Pooling Max Pooling is a pooling operation that calculates the maximum value for patches of a feature map, and uses it to create a downsampled (pooled) feature map. It is usually…

Local Interpretable Model-agnostic Explanations of Bayesian Predictive Models via Kullback-Leibler Projections

2018-10-05 · Tomi Peltola

We introduce a method, KL-LIME, for explaining predictions of Bayesian predictive models by projecting the information in the predictive distribution locally to a simpler, interpretable explanation model. The proposed ap…

Variable Selection

Adversarial Robustness Object Object Recognition

Convolutional Dynamic Alignment Networks for Interpretable Classifications

Code (1)

Methods 이 논문이 사용한 방법론

Similar Papers 제목 키워드 기반

Local Interpretable Model-agnostic Explanations of Bayesian Predictive Models via Kullback-Leibler Projections

Optimising for Interpretability: Convolutional Dynamic Alignment Networks

Agentic Moderation: Multi-Agent Design for Safer Vision-Language Models

Extracting Interpretable Concept-Based Decision Trees from CNNs

Interpretable Graph Capsule Networks for Object Recognition