The reasonable definition of semantic interpretability presents the core challenge in explainable AI. This paper proposes a method to modify a traditional convolutional neural network (CNN) into an interpretable compositional CNN, in order to learn filters that encode meaningful visual patterns in intermediate convolutional layers. In a compositional CNN, each filter is supposed to consistently represent a specific compositional object part or image region with a clear meaning. The compositional CNN learns from image labels for classification without any annotations of parts or regions for supervision. Our method can be broadly applied to different types of CNNs. Experiments have demonstrated the effectiveness of our method.
There are two main lines of research on visual question answering (VQA): compositional model with explicit multi-hop reasoning, and monolithic network with implicit reasoning in the latent feature space. The former excels in interpretability and compositionality but fails on real-world images, while the latter usually achieves better performance due to model flexibility and parameter efficiency. We aim to combine the two to build an interpretable framework for real-world compositional VQA. In our framework, images and questions are disentangled into scene graphs and programs, and a symbolic pr…
Computer vision systems in real-world applications need to be robust to partial occlusion while also being explainable. In this work, we show that black-box deep convolutional neural networks (DCNNs) have only limited robustness to partial occlusion. We overcome these limitations by unifying DCNNs with part-based models into Compositional Convolutional Neural Networks (CompositionalNets) - an interpretable deep architecture with innate robustness to partial occlusion. Specifically, we propose to replace the fully connected classification head of DCNNs with a differentiable compositional model …
Compositional convolutional networks are generative compositional models of neural network features, that achieve state of the art results when classifying partially occluded objects, even when they have not been exposed to occluded objects during training. In this work, we study the performance of CompositionalNets at localizing occluders in images. We show that the original model is not able to localize occluders well. We propose to overcome this limitation by modeling the feature activations as a mixture of von-Mises-Fisher distributions, which also allows for an end-to-end training of Comp…
Recent findings show that deep convolutional neural networks (DCNNs) do not generalize well under partial occlusion. Inspired by the success of compositional models at classifying partially occluded objects, we propose to integrate compositional models and DCNNs into a unified deep model with innate robustness to partial occlusion. We term this architecture Compositional Convolutional Neural Network. In particular, we propose to replace the fully connected classification head of a DCNN with a differentiable compositional model. The generative nature of the compositional model enables it to loc…
Johns Hopkins University（约翰斯·霍普金斯大学）；Max Planck Institute for Informatics（马克斯·普朗克信息学研究所）
The field of computational pathology has witnessed great advancements since deep neural networks have been widely applied. These networks usually require large numbers of annotated data to train vast parameters. However, it takes significant effort to annotate a large histopathology dataset. We introduce a light-weight and interpretable model for nuclei detection and weakly-supervised segmentation. It only requires annotations on isolated nucleus, rather than on all nuclei in the dataset. Besides, it is a generative compositional model that first locates parts of nucleus, then learns the spati…
Tokyo Institute of Technology（东京工业大学）；Kadinche Corporation（Kadinche公司）；National Institute of Advanced Industrial Science and Technology（国立先进工业科学技术研究所）
Image restoration from a single image degradation type, such as blurring, hazing, random noise, and compression has been investigated for decades. However, image degradations in practice are often a mixture of several types of degradation. Such compositional degradations complicate restoration because they require the differentiation of different degradation types and levels. In this paper, we propose a convolutional neural network (CNN) model for estimating the degradation properties of a given degraded image. Furthermore, we introduce an image restoration CNN model that adopts the estimated …
University of California, Los Angeles（加利福尼亚大学洛杉矶分校）
This paper proposes a method to modify traditional convolutional neural networks (CNNs) into interpretable CNNs, in order to clarify knowledge representations in high conv-layers of CNNs. In an interpretable CNN, each filter in a high conv-layer represents a certain object part. We do not need any annotations of object parts or textures to supervise the learning process. Instead, the interpretable CNN automatically assigns each filter in a high conv-layer with an object part during the learning process. Our method can be applied to different types of CNNs with different structures. The clear k…
The model parameters of convolutional neural networks (CNNs) are determined by backpropagation (BP). In this work, we propose an interpretable feedforward (FF) design without any BP as a reference. The FF design adopts a data-centric approach. It derives network parameters of the current layer based on data statistics from the output of the previous layer in a one-pass manner. To construct convolutional layers, we develop a new signal transform, called the Saab (Subspace Approximation with Adjusted Bias) transform. It is a variant of the principal component analysis (PCA) with an added bias ve…
Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）；University of Chinese Academy of Sciences（中国科学院大学）
Convolutional neural networks (ConvNets) are widely used in real life. People usually use ConvNets which pre-trained on a fixed number of classes. However, for different application scenarios, we usually do not need all of the classes, which means ConvNets are redundant when dealing with these tasks. This paper focuses on the redundancy of ConvNet channels. We proposed a novel idea: using an interpretable manner to find the most important channels for every single class (dissect), and dynamically run channels according to classes in need (reconstruct). For VGG16 pre-trained on CIFAR-10, we onl…
The discriminative power of modern deep learning models for 3D human action recognition is growing ever so potent. In conjunction with the recent resurgence of 3D human action representation with 3D skeletons, the quality and the pace of recent progress have been significant. However, the inner workings of state-of-the-art learning based methods in 3D human action recognition still remain mostly black-box. In this work, we propose to use a new class of models known as Temporal Convolutional Neural Networks (TCN) for 3D human action recognition. Compared to popular LSTM-based Recurrent Neural N…
Compositional Factorization of Visual Scenes with Convolutional Sparse Coding and Resonator Networks
Redwood Center for Theoretical Neuroscience, UC Berkeley（加州大学伯克利分校红木理论神经科学中心）；Université Paris-Saclay（巴黎-萨克雷大学）；Örebro University（厄勒布鲁大学）；Research Institutes of Sweden（瑞典研究院）
University of California, Berkeley（加州大学伯克利分校）；Boston University（波士顿大学）
In complex inferential tasks like question answering, machine learning models must confront two challenges: the need to implement a compositional reasoning process, and, in many applications, the need for this reasoning process to be interpretable to assist users in both development and prediction. Existing models designed to produce interpretable traces of their decision-making process typically require these traces to be supervised at training time. In this paper, we present a novel neural modular approach that performs compositional reasoning by automatically inducing a desired sub-task dec…
Sun Yat-sen University（中山大学）；DMAI Great China（DMAI大中华区）
Collaborative reasoning for understanding image-question pairs is a very critical but underexplored topic in interpretable visual question answering systems. Although very recent studies have attempted to use explicit compositional processes to assemble multiple subtasks embedded in questions, their models heavily rely on annotations or handcrafted rules to obtain valid reasoning processes, which leads to either heavy workloads or poor performance on compositional reasoning. In this paper, to better align image and language domains in diverse and unrestricted cases, we propose a novel neural n…
This work proposes an interpretable multi-view deep neural network architecture, namely optimal discriminant multi-view tensor convolutional network (ODMTCNet), by integrating statistical machine learning (SML) principles with the deep neural network (DNN) architecture.
Wellcome / EPSRC Centre for Interventional and Surgical Sciences（威康/EPSRC介入与外科科学中心）；University College London Hospitals（伦敦大学学院医院）；Gastroenterology National Taiwan University（台湾大学胃肠病学科）；E-Da Cancer Hospital（义大癌症医院）；Innersight Labs（内视实验室）
In this work, we have concentrated our efforts on the interpretability of classification results coming from a fully convolutional neural network. Motivated by the classification of oesophageal tissue for real-time detection of early squamous neoplasia, the most frequent kind of oesophageal cancer in Asia, we present a new dataset and a novel deep learning method that by means of deep supervision and a newly introduced concept, the embedded Class Activation Map (eCAM), focuses on the interpretability of results as a design constraint of a convolutional network. We present a new approach to vis…
University of Surrey（萨里大学）；Adobe Research（Adobe研究院）
We present an algorithm for searching image collections using free-hand sketches that describe the appearance and relative positions of multiple objects. Sketch based image retrieval (SBIR) methods predominantly match queries containing a single, dominant object invariant to its position within an image. Our work exploits drawings as a concise and intuitive representation for specifying entire scene compositions. We train a convolutional neural network (CNN) to encode masked visual features from sketched objects, pooling these into a spatial descriptor encoding the spatial relationships and ap…
University of Birmingham（伯明翰大学）；University of Ljubljana（卢布尔雅那大学）
Hierarchical feature learning based on convolutional neural networks (CNN) has recently shown significant potential in various computer vision tasks. While allowing high-quality discriminative feature learning, the downside of CNNs is the lack of explicit structure in features, which often leads to overfitting, absence of reconstruction from partial observations and limited generative abilities. Explicit structure is inherent in hierarchical compositional models, however, these lack the ability to optimize a well-defined cost function. We propose a novel analytic model of a basic unit in a lay…
University of Southern California（南加州大学）；US Army Research Laboratory（美国陆军研究实验室）
In this work, we study the power of Saak features as an effort towards interpretable deep learning. Being inspired by the operations of convolutional layers of convolutional neural networks, multi-stage Saak transform was proposed. Based on this foundation, we provide an in-depth examination on Saak features, which are coefficients of the Saak transform, by analyzing their properties through visualization and demonstrating their applications in image classification. Being similar to CNN features, Saak features at later stages have larger receptive fields, yet they are obtained in a one-pass fe…
Princeton University（普林斯顿大学）；The Chinese University of Hong Kong（香港中文大学）；Microsoft（微软）；UCLA（加州大学洛杉矶分校）；Beth Israel Deaconess Medical Center（贝斯以色列女执事医疗中心）；Microsoft Research New England（微软新英格兰研究院）；VSCO
We propose DeepMiner, a framework to discover interpretable representations in deep neural networks and to build explanations for medical predictions. By probing convolutional neural networks (CNNs) trained to classify cancer in mammograms, we show that many individual units in the final convolutional layer of a CNN respond strongly to diseased tissue concepts specified by the BI-RADS lexicon. After expert annotation of the interpretable units, our proposed method is able to generate explanations for CNN mammogram classification that are consistent with ground truth radiology reports on the Di…
Technical University of Denmark（丹麦技术大学）；Danish Research Centre for Magnetic Resonance（丹麦磁共振研究中心）
Sleep studies are important for diagnosing sleep disorders such as insomnia, narcolepsy or sleep apnea. They rely on manual scoring of sleep stages from raw polisomnography signals, which is a tedious visual task requiring the workload of highly trained professionals. Consequently, research efforts to purse for an automatic stage scoring based on machine learning techniques have been carried out over the last years. In this work, we resort to multitaper spectral analysis to create visually interpretable images of sleep patterns from EEG signals as inputs to a deep convolutional network trained…
University of California, Los Angeles（加利福尼亚大学洛杉矶分校）
While deep learning methods are increasingly being applied to tasks such as computer-aided diagnosis, these models are difficult to interpret, do not incorporate prior domain knowledge, and are often considered as a "black-box." The lack of model interpretability hinders them from being fully understood by target users such as radiologists. In this paper, we present a novel interpretable deep hierarchical semantic convolutional neural network (HSCNN) to predict whether a given pulmonary nodule observed on a computed tomography (CT) scan is malignant. Our network provides two levels of output: …
Bosch Center for Artificial Intelligence (BCAI)（博世人工智能中心）；University of Bonn（波恩大学）
To verify and validate networks, it is essential to gain insight into their decisions, limitations as well as possible shortcomings of training data. In this work, we propose a post-hoc, optimization based visual explanation method, which highlights the evidence in the input image for a specific prediction. Our approach is based on a novel technique to defend against adversarial evidence (i.e. faulty evidence due to artefacts) by filtering gradients during optimization. The defense does not depend on human-tuned parameters. It enables explanations which are both fine-grained and preserve the c…
A Hybrid Fully Convolutional CNN-Transformer Model for Inherently Interpretable Disease Detection from Retinal Fundus Images
Hertie Institute for AI in Brain Health, University of Tübingen, Germany（人工智能与脑健康赫尔特研究所，图宾根大学，德国）；Tübingen AI Center, University of Tübingen, Germany（图宾根人工智能中心，图宾根大学，德国）
Technical University of Munich（慕尼黑工业大学）；Sharif University of Technology（谢里夫理工大学）；Johns Hopkins University（约翰斯·霍普金斯大学）
Interpretability in Graph Convolutional Networks (GCNs) has been explored to some extent in computer vision in general, yet, in the medical domain, it requires further examination. Moreover, most of the interpretability approaches for GCNs, especially in the medical domain, focus on interpreting the model in a post hoc fashion. In this paper, we propose an interpretable graph learning-based model which 1) interprets the clinical relevance of the input features towards the task, 2) uses the explanation to improve the model performance and, 3) learns a population level latent graph that may be u…
Learning compact and interpretable representations is a very natural task, which has not been solved satisfactorily even for simple binary datasets. In this paper, we review various ways of composing experts for binary data and argue that competitive forms of interaction are best suited to learn low-dimensional representations. We propose a new composition rule that discourages experts from focusing on similar structures and that penalizes opposing votes strongly so that abstaining from voting becomes more attractive. We also introduce a novel sequential initialization procedure, which is base…
We propose Recognition as Part Composition (RPC), an image encoding approach inspired by human cognition. It is based on the cognitive theory that humans recognize complex objects by components, and that they build a small compact vocabulary of concepts to represent each instance with. RPC encodes images by first decomposing them into salient parts, and then encoding each part as a mixture of a small number of prototypes, each representing a certain concept. We find that this type of learning inspired by human cognition can overcome hurdles faced by deep convolutional networks in low-shot gene…
Convolutional neural networks (CNNs) have shown great success in computer vision, approaching human-level performance when trained for specific tasks via application-specific loss functions. In this paper, we propose a method for augmenting and training CNNs so that their learned features are compositional. It encourages networks to form representations that disentangle objects from their surroundings and from each other, thereby promoting better generalization. Our method is agnostic to the specific details of the underlying CNN to which it is applied and can in principle be used with any CNN…
Lehigh University（里海大学）；University of Pennsylvania（宾夕法尼亚大学）
Identification of brain regions related to the specific neurological disorders are of great importance for biomarker and diagnostic studies. In this paper, we propose an interpretable Graph Convolutional Network (GCN) framework for the identification and classification of Alzheimer's disease (AD) using multi-modality brain imaging data. Specifically, we extended the Gradient Class Activation Mapping (Grad-CAM) technique to quantify the most discriminative features identified by GCN from brain connectivity patterns. We then utilized them to find signature regions of interest (ROIs) by detecting…
Tsinghua University（清华大学）；Tsinghua SIGS（清华大学深圳国际研究生院）；Peng Cheng Lab（鹏城实验室）；ByteDance AI Lab（字节跳动人工智能实验室）；University of Southern California（南加州大学）
Convolutional neural networks (CNNs) have been successfully used in a range of tasks. However, CNNs are often viewed as "black-box" and lack of interpretability. One main reason is due to the filter-class entanglement -- an intricate many-to-many correspondence between filters and classes. Most existing works attempt post-hoc interpretation on a pre-trained model, while neglecting to reduce the entanglement underlying the model. In contrast, we focus on alleviating filter-class entanglement during training. Inspired by cellular differentiation, we propose a novel strategy to train interpretabl…
University of Tübingen（蒂宾根大学）；Max Planck Institute for Informatics（马克斯·普朗克信息学研究所）；University of Amsterdam（阿姆斯特丹大学）
Learning a common representation space between vision and language allows deep networks to relate objects in the image to the corresponding semantic meaning. We present a model that learns a shared Gaussian mixture representation imposing the compositionality of the text onto the visual domain without having explicit location supervision. By combining the spatial transformer with a representation learning approach we learn to split images into separately encoded patches to associate visual and textual representations in an interpretable manner. On variations of MNIST and CIFAR10, our model is …
We present VISPROG, a neuro-symbolic approach to solving complex and compositional visual tasks given natural language instructions. VISPROG avoids the need for any task-specific training. Instead, it uses the in-context learning ability of large language models to generate python-like modular programs, which are then executed to get both the solution and a comprehensive and interpretable rationale. Each line of the generated program may invoke one of several off-the-shelf computer vision models, image processing routines, or python functions to produce intermediate outputs that may be consume…
In image captioning where fluency is an important factor in evaluation, e.g., $n$-gram metrics, sequential models are commonly used; however, sequential models generally result in overgeneralized expressions that lack the details that may be present in an input image. Inspired by the idea of the compositional neural module networks in the visual question answering task, we introduce a hierarchical framework for image captioning that explores both compositionality and sequentiality of natural language. Our algorithm learns to compose a detail-rich sentence by selectively attending to different …
The University of Texas at Austin（德克萨斯大学奥斯汀分校）；Lockheed Martin Corporation（洛克希德·马丁公司）
We propose an image-classification method to predict the perceived-relevance of text documents from eye-movements. An eye-tracking study was conducted where participants read short news articles, and rated them as relevant or irrelevant for answering a trigger question. We encode participants' eye-movement scanpaths as images, and then train a convolutional neural network classifier using these scanpath images. The trained classifier is used to predict participants' perceived-relevance of news articles from the corresponding scanpath images. This method is content-independent, as the classifie…
Shanghai Jiao Tong University（上海交通大学）；University of California, Los Angeles（加州大学洛杉矶分校）
In this paper, we present a method to mine object-part patterns from conv-layers of a pre-trained convolutional neural network (CNN). The mined object-part patterns are organized by an And-Or graph (AOG). This interpretable AOG representation consists of a four-layer semantic hierarchy, i.e., semantic parts, part templates, latent patterns, and neural units. The AOG associates each object part with certain neural units in feature maps of conv-layers. The AOG is constructed in a weakly-supervised manner, i.e., very few annotations (e.g., 3-20) of object parts are used to guide the learning of A…
Compositional structures between parts and objects are inherent in natural scenes. Modeling such compositional hierarchies via unsupervised learning can bring various benefits such as interpretability and transferability, which are important in many downstream tasks. In this paper, we propose the first deep latent variable model, called RICH, for learning Representation of Interpretable Compositional Hierarchies. At the core of RICH is a latent scene graph representation that organizes the entities of a scene into a tree structure according to their compositional relationships. During inferenc…
Johns Hopkins Department of Radiology and Radiological Sciences（约翰斯·霍普金斯大学放射学与放射科学系）
We formally demonstrate that the relative seriality model of Kallman, et al. maps exactly onto a simple type of convolutional neural network. This approach leads to a natural interpretation of feedforward connections in the convolutional layer and stacked intermediate pooling layers in terms of bystander effects and hierarchical tissue organization, respectively. These results serve as proof-of-principle for radiobiologically interpretable deep learning of normal tissue complication probability using large-scale imaging and dosimetry datasets.
Harbin Engineering University（哈尔滨工程大学）；NC State University（北卡罗来纳州立大学）；University of California, Los Angeles（加利福尼亚大学洛杉矶分校）
This paper proposes to learn hierarchical compositional AND-OR model for interpretable image synthesis by sparsifying the generator network. The proposed method adopts the scene-objects-parts-subparts-primitives hierarchy in image representation. A scene has different types (i.e., OR) each of which consists of a number of objects (i.e., AND). This can be recursively formulated across the scene-objects-parts-subparts hierarchy and is terminated at the primitive level (e.g., wavelets-like basis). To realize this AND-OR hierarchy in image synthesis, we learn a generator network that consists of t…
Shanghai Jiao Tong University（上海交通大学）；University of California, Los Angeles（加州大学洛杉矶分校）
This paper proposes a generic method to learn interpretable convolutional filters in a deep convolutional neural network (CNN) for object classification, where each interpretable filter encodes features of a specific object part. Our method does not require additional annotations of object parts or textures for supervision. Instead, we use the same training data as traditional CNNs. Our method automatically assigns each interpretable filter in a high conv-layer with an object part of a certain category during the learning process. Such explicit knowledge representations in conv-layers of CNN h…
Interpretable 3D Multi-Modal Residual Convolutional Neural Network for Mild Traumatic Brain Injury Diagnosis
School of Electrical Engineering and Computer Science, University of Queensland（昆士兰大学电气工程与计算机科学学院）；The Centre for Advanced Imaging, University of Queensland（昆士兰大学高级成像中心）；Australian Institute for Bioengineering and Nanotechnology, University of Queensland（昆士兰大学生物工程与纳米技术研究所）；ARC Training Centre for Innovation in Biomedical Imaging Technology（ARC生物医学成像技术创新培训中心）
Durham University（杜伦大学）；University of Glasgow（格拉斯哥大学）
Early prediction is clinically considered one of the essential parts of cerebral palsy (CP) treatment. We propose to implement a low-cost and interpretable classification system for supporting CP prediction based on General Movement Assessment (GMA). We design a Pytorch-based attention-informed graph convolutional network to early identify infants at risk of CP from skeletal data extracted from RGB videos. We also design a frequency-binning module for learning the CP movements in the frequency domain while filtering noise. Our system only requires consumer-grade RGB videos for training to supp…
GL-ICNN: An End-To-End Interpretable Convolutional Neural Network for the Diagnosis and Prediction of Alzheimer's Disease
Erasmus MC（伊拉斯姆斯医学中心）；University Medical Center Groningen（格罗宁根大学医学中心）；University Medical Center Utrecht（乌得勒支大学医学中心）；Radboud University Medical Center（拉德堡德大学医学中心）；Leiden University Medical Center（莱顿大学医学中心）；Amsterdam University Medical Center（阿姆斯特丹大学医学中心）；Maastricht University Medical Center（马斯特里赫特大学医学中心）
Tencent Jarvis Lab（腾讯天衍实验室）；Xi’an Jiaotong University（西安交通大学）；Peng Cheng Laboratory（鹏城实验室）；Macau University of Science and Technology（澳门科技大学）
As a common weather, rain streaks adversely degrade the image quality. Hence, removing rains from an image has become an important issue in the field. To handle such an ill-posed single image deraining task, in this paper, we specifically build a novel deep architecture, called rain convolutional dictionary network (RCDNet), which embeds the intrinsic priors of rain streaks and has clear interpretability. In specific, we first establish a RCD model for representing rain streaks and utilize the proximal gradient descent technique to design an iterative algorithm only containing simple operators…
As new data-sets for real-world visual reasoning and compositional question answering are emerging, it might be needed to use the visual feature extraction as a end-to-end process during training. This small contribution aims to suggest new ideas to improve the visual processing of traditional convolutional network for visual question answering (VQA). In this paper, we propose to modulate by a linguistic input a CNN augmented with self-attention. We show encouraging relative improvements for future research in this direction.
Center for Data Science（数据科学中心）；New York University（纽约大学）；Center for Neural Science（神经科学中心）；Howard Hughes Medical Institute（霍华德·休斯医学研究所）；Courant Inst. of Mathematical Sciences（库朗数学科学研究所）
Deep convolutional networks often append additive constant ("bias") terms to their convolution operations, enabling a richer repertoire of functional mappings. Biases are also used to facilitate training, by subtracting mean response over batches of training images (a component of "batch normalization"). Recent state-of-the-art blind denoising methods (e.g., DnCNN) seem to require these terms for their success. Here, however, we show that these networks systematically overfit the noise levels for which they are trained: when deployed at noise levels outside the training range, performance degr…
Convolutional neural network (CNN), with ability of feature learning and nonlinear mapping, has demonstrated its effectiveness in prognostics and health management (PHM). However, explanation on the physical meaning of a CNN architecture has rarely been studied. In this paper, a novel wavelet driven deep neural network termed as WaveletKernelNet (WKN) is presented, where a continuous wavelet convolutional (CWConv) layer is designed to replace the first convolutional layer of the standard CNN. This enables the first CWConv layer to discover more meaningful filters. Furthermore, only the scale p…
