Traditional saliency map methods, popularized in computer vision, highlight individual points (pixels) of the input that contribute the most to the model's output. However, in time-series they offer limited insights as semantically meaningful features are often found in other domains. We introduce Cross-domain Integrated Gradients, a generalization of Integrated Gradients. Our method enables feature attributions on any domain that can be formulated as an invertible, differentiable transformation of the time domain. Crucially, our derivation extends the original Integrated Gradients into the complex domain, enabling frequency-based attributions. We provide the necessary theoretical guarantees, namely, path independence and completeness. Our approach reveals interpretable, problem-specific attributions that time-domain methods cannot capture, on three real-world tasks: wearable sensor heart rate extraction, electroencephalography-based seizure detection, and zero-shot time-series forecasting. We release an open-source Tensorflow/PyTorch library to enable plug-and-play cross-domain explainability for time-series models. These results demonstrate the ability of cross-domain integrated gradients to provide semantically meaningful insights in time-series models that are impossible with traditional time-domain saliency.
Diffusion models have demonstrated impressive capabilities in synthesizing diverse content. However, despite their high-quality outputs, these models often perpetuate social biases, including those related to gender and race. These biases can potentially contribute to harmful real-world consequences, reinforcing stereotypes and exacerbating inequalities in various social contexts. While existing research on diffusion bias mitigation has predominantly focused on guiding content generation, it often neglects the intrinsic mechanisms within diffusion models that causally drive biased outputs. In this paper, we investigate the internal processes of diffusion models, identifying specific decision-making mechanisms, termed bias features, embedded within the model architecture. By directly manipulating these features, our method precisely isolates and adjusts the elements responsible for bias generation, permitting granular control over the bias levels in the generated content. Through experiments on both unconditional and conditional diffusion models across various social bias attributes, we demonstrate our method's efficacy in managing generation distribution while preserving image quality. We also dissect the discovered model mechanism, revealing different intrinsic features controlling fine-grained aspects of generation, boosting further research on mechanistic interpretability of diffusion models.
Histopathological image analysis is crucial for accurate cancer diagnosis and treatment planning. While deep learning models, especially convolutional neural networks, have advanced this field, their "black-box" nature raises concerns about interpretability and trustworthiness. Explainable Artificial Intelligence (XAI) techniques aim to address these concerns, but evaluating their effectiveness remains challenging. A significant issue with current occlusion-based XAI methods is that they often generate Out-of-Distribution (OoD) samples, leading to inaccurate evaluations. In this paper, we introduce Inpainting-Based Occlusion (IBO), a novel occlusion strategy that utilizes a Denoising Diffusion Probabilistic Model to inpaint occluded regions in histopathological images. By replacing cancerous areas with realistic, non-cancerous tissue, IBO minimizes OoD artifacts and preserves data integrity. We evaluate our method on the CAMELYON16 dataset through two phases: first, by assessing perceptual similarity using the Learned Perceptual Image Patch Similarity (LPIPS) metric, and second, by quantifying the impact on model predictions through Area Under the Curve (AUC) analysis. Our results demonstrate that IBO significantly improves perceptual fidelity, achieving nearly twice the improvement in LPIPS scores compared to the best existing occlusion strategy. Additionally, IBO increased the precision of XAI performance prediction from 42% to 71% compared to traditional methods. These results demonstrate IBO's potential to provide more reliable evaluations of XAI techniques, benefiting histopathology and other applications. The source code for this study is available at https://github.com/a-fsh-r/IBO.
Language models (LMs) have proven to be powerful tools for psycholinguistic research, but most prior work has focused on purely behavioural measures (e.g., surprisal comparisons). At the same time, research in model interpretability has begun to illuminate the abstract causal mechanisms shaping LM behavior. To help bring these strands of research closer together, we introduce CausalGym. We adapt and expand the SyntaxGym suite of tasks to benchmark the ability of interpretability methods to causally affect model behaviour. To illustrate how CausalGym can be used, we study the pythia models (14M--6.9B) and assess the causal efficacy of a wide range of interpretability methods, including linear probing and distributed alignment search (DAS). We find that DAS outperforms the other methods, and so we use it to study the learning trajectory of two difficult linguistic phenomena in pythia-1b: negative polarity item licensing and filler--gap dependencies. Our analysis shows that the mechanism implementing both of these tasks is learned in discrete stages, not gradually.
Image attribution algorithms aim to identify important regions that are highly relevant to model decisions. Although existing attribution solutions can effectively assign importance to target elements, they still face the following challenges: 1) existing attribution methods generate inaccurate small regions thus misleading the direction of correct attribution, and 2) the model cannot produce good attribution results for samples with wrong predictions. To address the above challenges, this paper re-models the above image attribution problem as a submodular subset selection problem, aiming to enhance model interpretability using fewer regions. To address the lack of attention to local regions, we construct a novel submodular function to discover more accurate small interpretation regions. To enhance the attribution effect for all samples, we also impose four different constraints on the selection of sub-regions, i.e., confidence, effectiveness, consistency, and collaboration scores, to assess the importance of various subsets. Moreover, our theoretical analysis substantiates that the proposed function is in fact submodular. Extensive experiments show that the proposed method outperforms SOTA methods on two face datasets (Celeb-A and VGG-Face2) and one fine-grained dataset (CUB-200-2011). For correctly predicted samples, the proposed method improves the Deletion and Insertion scores with an average of 4.9% and 2.5% gain relative to HSIC-Attribution. For incorrectly predicted samples, our method achieves gains of 81.0% and 18.4% compared to the HSIC-Attribution algorithm in the average highest confidence and Insertion score respectively. The code is released at https://github.com/RuoyuChen10/SMDL-Attribution.
In Federated Learning, clients train models on local data and send updates to a central server, which aggregates them into a global model using a fusion algorithm. This collaborative yet privacy-preserving training comes at a cost. FL developers face significant challenges in attributing global model predictions to specific clients. Localizing responsible clients is a crucial step towards (a) excluding clients primarily responsible for incorrect predictions and (b) encouraging clients who contributed high-quality models to continue participating in the future. Existing ML debugging approaches are inherently inapplicable as they are designed for single-model, centralized training. We introduce TraceFL, a fine-grained neuron provenance capturing mechanism that identifies clients responsible for a global model's prediction by tracking the flow of information from individual clients to the global model. Since inference on different inputs activates a different set of neurons of the global model, TraceFL dynamically quantifies the significance of the global model's neurons in a given prediction, identifying the most crucial neurons in the global model. It then maps them to the corresponding neurons in every participating client to determine each client's contribution, ultimately localizing the responsible client. We evaluate TraceFL on six datasets, including two real-world medical imaging datasets and four neural networks, including advanced models such as GPT. TraceFL achieves 99% accuracy in localizing the responsible client in FL tasks spanning both image and text classification tasks. At a time when state-of-the-artML debugging approaches are mostly domain-specific (e.g., image classification only), TraceFL is the first technique to enable highly accurate automated reasoning across a wide range of FL applications.
State-of-the-art deep neural networks (DNNs) are highly effective at tackling many real-world tasks. However, their wide adoption in mission-critical contexts is hampered by two major weaknesses - their susceptibility to adversarial attacks and their opaqueness. The former raises concerns about the security and generalization of DNNs in real-world conditions, whereas the latter impedes users' trust in their output. In this research, we (1) examine the effect of adversarial robustness on interpretability and (2) present a novel approach for improving the interpretability of DNNs that is based on regularization of neural activation sensitivity. We evaluate the interpretability of models trained using our method to that of standard models and models trained using state-of-the-art adversarial robustness techniques. Our results show that adversarially robust models are superior to standard models and that models trained using our proposed method are even better than adversarially robust models in terms of interpretability.
This paper presents a new efficient black-box attribution method based on Hilbert-Schmidt Independence Criterion (HSIC), a dependence measure based on Reproducing Kernel Hilbert Spaces (RKHS). HSIC measures the dependence between regions of an input image and the output of a model based on kernel embeddings of distributions. It thus provides explanations enriched by RKHS representation capabilities. HSIC can be estimated very efficiently, significantly reducing the computational cost compared to other black-box attribution methods. Our experiments show that HSIC is up to 8 times faster than the previous best black-box attribution methods while being as faithful. Indeed, we improve or match the state-of-the-art of both black-box and white-box attribution methods for several fidelity metrics on Imagenet with various recent model architectures. Importantly, we show that these advances can be transposed to efficiently and faithfully explain object detection models such as YOLOv4. Finally, we extend the traditional attribution methods by proposing a new kernel enabling an ANOVA-like orthogonal decomposition of importance scores based on HSIC, allowing us to evaluate not only the importance of each image patch but also the importance of their pairwise interactions. Our implementation is available at https://github.com/paulnovello/HSIC-Attribution-Method.
We propose a new method for spatio-temporal forecasting on arbitrarily distributed points. Assuming that the observed system follows an unknown partial differential equation, we derive a continuous-time model for the dynamics of the data via the finite element method. The resulting graph neural network estimates the instantaneous effects of the unknown dynamics on each cell in a meshing of the spatial domain. Our model can incorporate prior knowledge via assumptions on the form of the unknown PDE, which induce a structural bias towards learning specific processes. Through this mechanism, we derive a transport variant of our model from the convection equation and show that it improves the transfer performance to higher-resolution meshes on sea surface temperature and gas flow forecasting against baseline models representing a selection of spatio-temporal forecasting methods. A qualitative analysis shows that our model disentangles the data dynamics into their constituent parts, which makes it uniquely interpretable.
The COVID-19 pandemic has significantly impacted the tourism and hospitality sector. Public policies such as travel restrictions and stay-at-home orders had significantly affected tourist activities and service businesses' operations and profitability. To this end, it is essential to develop an interpretable forecast model that supports managerial and organizational decision-making. We developed DemandNet, a novel deep learning framework for predicting time series data under the influence of the COVID-19 pandemic. The framework starts by selecting the top static and dynamic features embedded in the time series data. Then, it includes a nonlinear model which can provide interpretable insight into the previously seen data. Lastly, a prediction model is developed to leverage the above characteristics to make robust long-term forecasts. We evaluated the framework using daily hotel demand and revenue data from eight cities in the US. Our findings reveal that DemandNet outperforms the state-of-art models and can accurately predict the impact of the COVID-19 pandemic on hotel demand and revenues.
Deep learning techniques have recently brought many improvements in the field of neural network training, especially for prognosis and health management. The success of such an intelligent health assessment model depends not only on the availability of labeled historical data but also on the careful samples selection. However, in real operating systems such as induction machines, which generally have a long reliable life, storing the entire operation history, including deterioration (i.e. bearings), will be very expensive and difficult to feed accurately into the training model. Other alternatives sequentially store samples that hold degradation patterns similar to real ones in damage behavior by imposing an accelerated deterioration. Labels lack and differences in distributions caused by the imposed deterioration will ultimately discriminate the training model and limit its knowledge capacity. In an attempt to overcome these drawbacks, a novel sequence-by-sequence deep learning algorithm able to expand the generalization capacity by transferring obtained knowledge from life cycles of similar systems is proposed. The new algorithm aims to determine health status by involving long short-term memory neural network as a primary component of adaptive learning to extract both health stage and health index inferences. Experimental validations are performed using the PRONOSTIA bearing degradation datasets.
Explaining deep learning model inferences is a promising venue for scientific understanding, improving safety, uncovering hidden biases, evaluating fairness, and beyond, as argued by many scholars. One of the principal benefits of counterfactual explanations is allowing users to explore "what-if" scenarios through what does not and cannot exist in the data, a quality that many other forms of explanation such as heatmaps and influence functions are inherently incapable of doing. However, most previous work on generative explainability cannot disentangle important concepts effectively, produces unrealistic examples, or fails to retain relevant information. We propose a novel approach, DISSECT, that jointly trains a generator, a discriminator, and a concept disentangler to overcome such challenges using little supervision. DISSECT generates Concept Traversals (CTs), defined as a sequence of generated examples with increasing degrees of concepts that influence a classifier's decision. By training a generative model from a classifier's signal, DISSECT offers a way to discover a classifier's inherent "notion" of distinct concepts automatically rather than rely on user-predefined concepts. We show that DISSECT produces CTs that (1) disentangle several concepts, (2) are influential to a classifier's decision and are coupled to its reasoning due to joint training (3), are realistic, (4) preserve relevant information, and (5) are stable across similar inputs. We validate DISSECT on several challenging synthetic and realistic datasets where previous methods fall short of satisfying desirable criteria for interpretability and show that it performs consistently well and better than existing methods. Finally, we present experiments showing applications of DISSECT for detecting potential biases of a classifier and identifying spurious artifacts that impact predictions.
Modern machine learning systems based on neural networks have shown great success in learning complex data patterns while being able to make good predictions on unseen data points. However, the limited interpretability of these systems hinders further progress and application to several domains in the real world. This predicament is exemplified by time consuming model selection and the difficulties faced in predictive explainability, especially in the presence of adversarial examples. In this paper, we take a step towards better understanding of neural networks by introducing a local polytope interpolation method. The proposed Deep Non Negative Kernel regression (NNK) interpolation framework is non parametric, theoretically simple and geometrically intuitive. We demonstrate instance based explainability for deep learning models and develop a method to identify models with good generalization properties using leave one out estimation. Finally, we draw a rationalization to adversarial and generative examples which are inevitable from an interpolation view of machine learning.
The outbreak of COVID-19 has shocked the entire world with its fairly rapid spread and has challenged different sectors. One of the most effective ways to limit its spread is the early and accurate diagnosing infected patients. Medical imaging, such as X-ray and Computed Tomography (CT), combined with the potential of Artificial Intelligence (AI), plays an essential role in supporting medical personnel in the diagnosis process. Thus, in this article five different deep learning models (ResNet18, ResNet34, InceptionV3, InceptionResNetV2 and DenseNet161) and their ensemble, using majority voting have been used to classify COVID-19, pneumoni{\ae} and healthy subjects using chest X-ray images. Multilabel classification was performed to predict multiple pathologies for each patient, if present. Firstly, the interpretability of each of the networks was thoroughly studied using local interpretability methods - occlusion, saliency, input X gradient, guided backpropagation, integrated gradients, and DeepLIFT, and using a global technique - neuron activation profiles. The mean Micro-F1 score of the models for COVID-19 classifications ranges from 0.66 to 0.875, and is 0.89 for the ensemble of the network models. The qualitative results showed that the ResNets were the most interpretable models. This research demonstrates the importance of using interpretability methods to compare different models before making a decision regarding the best performing model.
This paper explores interpretability techniques for two of the most successful learning algorithms in medical decision-making literature: deep neural networks and random forests. We applied these algorithms in a real-world medical dataset containing information about patients with cancer, where we learn models that try to predict the type of cancer of the patient, given their set of medical activity records. We explored different algorithms based on neural network architectures using long short term deep neural networks, and random forests. Since there is a growing need to provide decision-makers understandings about the logic of predictions of black boxes, we also explored different techniques that provide interpretations for these classifiers. In one of the techniques, we intercepted some hidden layers of these neural networks and used autoencoders in order to learn what is the representation of the input in the hidden layers. In another, we investigated an interpretable model locally around the random forest's prediction. Results show learning an interpretable model locally around the model's prediction leads to a higher understanding of why the algorithm is making some decision. Use of local and linear model helps identify the features used in prediction of a specific instance or data point. We see certain distinct features used for predictions that provide useful insights about the type of cancer, along with features that do not generalize well. In addition, the structured deep learning approach using autoencoders provided meaningful prediction insights, which resulted in the identification of nonlinear clusters correspondent to the patients' different types of cancer.
Deep neural network pruning and quantization techniques have demonstrated it is possible to achieve high levels of compression with surprisingly little degradation to test set accuracy. However, this measure of performance conceals significant differences in how different classes and images are impacted by model compression techniques. We find that models with radically different numbers of weights have comparable top-line performance metrics but diverge considerably in behavior on a narrow subset of the dataset. This small subset of data points, which we term Pruning Identified Exemplars (PIEs) are systematically more impacted by the introduction of sparsity. Compression disproportionately impacts model performance on the underrepresented long-tail of the data distribution. PIEs over-index on atypical or noisy images that are far more challenging for both humans and algorithms to classify. Our work provides intuition into the role of capacity in deep neural networks and the trade-offs incurred by compression. An understanding of this disparate impact is critical given the widespread deployment of compressed models in the wild.
Deep neural networks are being used increasingly to automate data analysis and decision making, yet their decision-making process is largely unclear and is difficult to explain to the end users. In this paper, we address the problem of Explainable AI for deep neural networks that take images as input and output a class probability. We propose an approach called RISE that generates an importance map indicating how salient each pixel is for the model's prediction. In contrast to white-box approaches that estimate pixel importance using gradients or other internal network state, RISE works on black-box models. It estimates importance empirically by probing the model with randomly masked versions of the input image and obtaining the corresponding outputs. We compare our approach to state-of-the-art importance extraction methods using both an automatic deletion/insertion metric and a pointing metric based on human-annotated object segments. Extensive experiments on several benchmark datasets show that our approach matches or exceeds the performance of other methods, including white-box approaches. Project page: http://cs-people.bu.edu/vpetsiuk/rise/
This paper addresses the visualisation of image classification models, learnt using deep Convolutional Networks (ConvNets). We consider two visualisation techniques, based on computing the gradient of the class score with respect to the input image. The first one generates an image, which maximises the class score [Erhan et al., 2009], thus visualising the notion of the class, captured by a ConvNet. The second technique computes a class saliency map, specific to a given image and class. We show that such maps can be employed for weakly supervised object segmentation using classification ConvNets. Finally, we establish the connection between the gradient-based ConvNet visualisation methods and deconvolutional networks [Zeiler et al., 2013].
Despite widespread adoption, machine learning models remain mostly black boxes. Understanding the reasons behind predictions is, however, quite important in assessing trust, which is fundamental if one plans to take action based on a prediction, or when choosing whether to deploy a new model. Such understanding also provides insights into the model, which can be used to transform an untrustworthy model or prediction into a trustworthy one. In this work, we propose LIME, a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner, by learning an interpretable model locally around the prediction. We also propose a method to explain models by presenting representative individual predictions and their explanations in a non-redundant way, framing the task as a submodular optimization problem. We demonstrate the flexibility of these methods by explaining different models for text (e.g. random forests) and image classification (e.g. neural networks). We show the utility of explanations via novel experiments, both simulated and with human subjects, on various scenarios that require trust: deciding if one should trust a prediction, choosing between models, improving an untrustworthy classifier, and identifying why a classifier should not be trusted.
We propose a technique for producing "visual explanations" for decisions from a large class of CNN-based models, making them more transparent. Our approach - Gradient-weighted Class Activation Mapping (Grad-CAM), uses the gradients of any target concept, flowing into the final convolutional layer to produce a coarse localization map highlighting important regions in the image for predicting the concept. Grad-CAM is applicable to a wide variety of CNN model-families: (1) CNNs with fully-connected layers, (2) CNNs used for structured outputs, (3) CNNs used in tasks with multimodal inputs or reinforcement learning, without any architectural changes or re-training. We combine Grad-CAM with fine-grained visualizations to create a high-resolution class-discriminative visualization and apply it to off-the-shelf image classification, captioning, and visual question answering (VQA) models, including ResNet-based architectures. In the context of image classification models, our visualizations (a) lend insights into their failure modes, (b) are robust to adversarial images, (c) outperform previous methods on localization, (d) are more faithful to the underlying model and (e) help achieve generalization by identifying dataset bias. For captioning and VQA, we show that even non-attention based models can localize inputs. We devise a way to identify important neurons through Grad-CAM and combine it with neuron names to provide textual explanations for model decisions. Finally, we design and conduct human studies to measure if Grad-CAM helps users establish appropriate trust in predictions from models and show that Grad-CAM helps untrained users successfully discern a 'stronger' nodel from a 'weaker' one even when both make identical predictions. Our code is available at https://github.com/ramprs/grad-cam/, along with a demo at http://gradcam.cloudcv.org, and a video at youtu.be/COjUB9Izk6E.