Interpretable Deep Learning Model Research Articles

Revealing the Impact of Genomic Alterations on Cancer Cell Signaling with an Interpretable Deep Learning Model.

Cancer is a disease of aberrant cellular signaling resulting from somatic genomic alterations (SGAs). Heterogeneous SGA events in tumors lead to tumor-specific signaling system aberrations. We interpret the cancer signaling system as a causal graphical model, where SGAs affect signaling proteins, propagate their effects through signal transduction, and ultimately change gene expression. To represent such a system, we developed a deep learning model called redundant-input neural network (RINN) with a transparent redundant-input architecture. Our findings demonstrate that by utilizing SGAs as inputs, the RINN can encode their impact on the signaling system and predict gene expression accurately when measured as the area under ROC curves. Moreover, the RINN can discover the shared functional impact (similar embeddings) of SGAs that perturb a common signaling pathway (e.g., PI3K, Nrf2, and TGF). Furthermore, the RINN exhibits the ability to discover known relationships in cellular signaling systems.

Assisting schizophrenia diagnosis using clinical electroencephalography and interpretable graph neural networks: a real-world and cross-site study.

Schizophrenia (SCZ) is a chronic and serious mental disorder with a high mortality rate. At present, there is a lack of objective, cost-effective and widely disseminated diagnosis tools to address this mental health crisis globally. Clinical electroencephalogram (EEG) is a noninvasive technique to measure brain activity with high temporal resolution, and accumulating evidence demonstrates that clinical EEG is capable of capturing abnormal SCZ neuropathology. Although EEG-based automated diagnostic tools have obtained impressive performance on individual datasets, the transportability of potential EEG biomarkers in cross-site real-world application is still an open question. To address the challenges of small sample sizes and population heterogeneity, we develop an advanced interpretable deep learning model using multimodal clinical EEG features and demographic information as inputs to graph neural networks, and further propose different transfer learning strategies to adapt to different clinical scenarios. Taking the disease discrimination of health control (HC) and SCZ with 1030 participants as a use case, our model is trained on a small clinical dataset (N = 188, Chinese) and enhanced using a large-scale public dataset (N = 508, American) of adult participants. Cross-site validation from an independent dataset of adult participants (N = 157, Chinese) produced stable performance, with AUCs of 0.793-0.852 and accuracies of 0.786-0.858 for different SCZ prevalence, respectively. In addition, cross-site validation from another dataset of adolescent boys (N = 84, Russian) yielded an AUC of 0.702 and an accuracy of 0.690. Moreover, feature visualization further revealed that the ranking of feature importance varied significantly among different datasets, and that EEG theta and alpha band power appeared to be the most significant and translational biomarkers of SCZ pathology. Overall, our promising results demonstrate the feasibility of SCZ discrimination using EEG biomarkers in multiple clinical settings.

Knowledge graph aids comprehensive explanation of drug and chemical toxicity.

In computational toxicology, prediction of complex endpoints has always been challenging, as they often involve multiple distinct mechanisms. State-of-the-art models are either limited by low accuracy, or lack of interpretability due to their black-box nature. Here, we introduce AIDTox, an interpretable deep learning model which incorporates curated knowledge of chemical-gene connections, gene-pathway annotations, and pathway hierarchy. AIDTox accurately predicts cytotoxicity outcomes in HepG2 and HEK293 cells. It also provides comprehensive explanations of cytotoxicity covering multiple aspects of drug activity, including target interaction, metabolism, and elimination. In summary, AIDTox provides a computational framework for unveiling cellular mechanisms for complex toxicity endpoints.

PINNet: a deep neural network with pathway prior knowledge for Alzheimer's disease.

Identification of Alzheimer's Disease (AD)-related transcriptomic signatures from blood is important for early diagnosis of the disease. Deep learning techniques are potent classifiers for AD diagnosis, but most have been unable to identify biomarkers because of their lack of interpretability. To address these challenges, we propose a pathway information-based neural network (PINNet) to predict AD patients and analyze blood and brain transcriptomic signatures using an interpretable deep learning model. PINNet is a deep neural network (DNN) model with pathway prior knowledge from either the Gene Ontology or Kyoto Encyclopedia of Genes and Genomes databases. Then, a backpropagation-based model interpretation method was applied to reveal essential pathways and genes for predicting AD. The performance of PINNet was compared with a DNN model without a pathway. Performances of PINNet outperformed or were similar to those of DNN without a pathway using blood and brain gene expressions, respectively. Moreover, PINNet considers more AD-related genes as essential features than DNN without a pathway in the learning process. Pathway analysis of protein-protein interaction modules of highly contributed genes showed that AD-related genes in blood were enriched with cell migration, PI3K-Akt, MAPK signaling, and apoptosis in blood. The pathways enriched in the brain module included cell migration, PI3K-Akt, MAPK signaling, apoptosis, protein ubiquitination, and t-cell activation. By integrating prior knowledge about pathways, PINNet can reveal essential pathways related to AD. The source codes are available at https://github.com/DMCB-GIST/PINNet.

The State of the Art in Deep Learning Applications, Challenges, and Future Prospects: A Comprehensive Review of Flood Forecasting and Management

Floods are a devastating natural calamity that may seriously harm both infrastructure and people. Accurate flood forecasts and control are essential to lessen these effects and safeguard populations. By utilizing its capacity to handle massive amounts of data and provide accurate forecasts, deep learning has emerged as a potent tool for improving flood prediction and control. The current state of deep learning applications in flood forecasting and management is thoroughly reviewed in this work. The review discusses a variety of subjects, such as the data sources utilized, the deep learning models used, and the assessment measures adopted to judge their efficacy. It assesses current approaches critically and points out their advantages and disadvantages. The article also examines challenges with data accessibility, the interpretability of deep learning models, and ethical considerations in flood prediction. The report also describes potential directions for deep-learning research to enhance flood predictions and control. Incorporating uncertainty estimates into forecasts, integrating many data sources, developing hybrid models that mix deep learning with other methodologies, and enhancing the interpretability of deep learning models are a few of these. These research goals can help deep learning models become more precise and effective, which will result in better flood control plans and forecasts. Overall, this review is a useful resource for academics and professionals working on the topic of flood forecasting and management. By reviewing the current state of the art, emphasizing difficulties, and outlining potential areas for future study, it lays a solid basis. Communities may better prepare for and lessen the destructive effects of floods by implementing cutting-edge deep learning algorithms, thereby protecting people and infrastructure.

A self-supervised vision transformer to predict survival from histopathology in renal cell carcinoma

PurposeTo develop and validate an interpretable deep learning model to predict overall and disease-specific survival (OS/DSS) in clear cell renal cell carcinoma (ccRCC).MethodsDigitised haematoxylin and eosin-stained slides from The Cancer Genome Atlas were used as a training set for a vision transformer (ViT) to extract image features with a self-supervised model called DINO (self-distillation with no labels). Extracted features were used in Cox regression models to prognosticate OS and DSS. Kaplan–Meier for univariable evaluation and Cox regression analyses for multivariable evaluation of the DINO-ViT risk groups were performed for prediction of OS and DSS. For validation, a cohort from a tertiary care centre was used.ResultsA significant risk stratification was achieved in univariable analysis for OS and DSS in the training (n = 443, log rank test, p < 0.01) and validation set (n = 266, p < 0.01). In multivariable analysis, including age, metastatic status, tumour size and grading, the DINO-ViT risk stratification was a significant predictor for OS (hazard ratio [HR] 3.03; 95%-confidence interval [95%-CI] 2.11–4.35; p < 0.01) and DSS (HR 4.90; 95%-CI 2.78–8.64; p < 0.01) in the training set but only for DSS in the validation set (HR 2.31; 95%-CI 1.15–4.65; p = 0.02). DINO-ViT visualisation showed that features were mainly extracted from nuclei, cytoplasm, and peritumoural stroma, demonstrating good interpretability.ConclusionThe DINO-ViT can identify high-risk patients using histological images of ccRCC. This model might improve individual risk-adapted renal cancer therapy in the future.

Interpretable deep learning architectures for improving drug response prediction performance: myth or reality?

Interpretable deep learning (DL) models that can provide biological insights, in addition to accurate predictions, are of great interest to the biomedical community. Recently, interpretable DL models that incorporate signaling pathways have been proposed for drug response prediction (DRP). While these models improve interpretability, it is unclear whether this comes at the cost of less accurate DRPs, or a prediction improvement can also be obtained. We comprehensively and systematically assessed four state-of-the-art interpretable DL models using three pathway collections to assess their ability in making accurate predictions on unseen samples from the same dataset, as well as their generalizability to an independent dataset. Our results showed that models that explicitly incorporate pathway information in the form of a latent layer perform worse compared to models that incorporate this information implicitly. However, in most evaluation setups, the best performance was achieved using a black-box multilayer perceptron, and the performance of a random forests baseline was comparable to those of the interpretable models. Replacing the signaling pathways with randomly generated pathways showed a comparable performance for the majority of the models. Finally, the performance of all models deteriorated when applied to an independent dataset. These results highlight the importance of systematic evaluation of newly proposed models using carefully selected baselines. We provide different evaluation setups and baseline models that can be used to achieve this goal. Implemented models and datasets are provided at https://doi.org/10.5281/zenodo.7787178 and https://doi.org/10.5281/zenodo.7101665, respectively.

Classification and structural characteristics of amorphous materials based on interpretable deep learning

Defining the structure characteristics of amorphous materials is one of the fundamental problems that need to be solved urgently in complex materials because of their complex structure and long-range disorder. In this study, we develop an interpretable deep learning model capable of accurately classifying amorphous configurations and characterizing their structural properties. The results demonstrate that the multi-dimensional hybrid convolutional neural network can classify the two-dimensional (2D) liquids and amorphous solids of molecular dynamics simulation. The classification process does not make a priori assumptions on the amorphous particle environment, and the accuracy is 92.75%, which is better than other convolutional neural networks. Moreover, our model utilizes the gradient-weighted activation-like mapping method, which generates activation-like heat maps that can precisely identify important structures in the amorphous configuration maps. We obtain an order parameter from the heatmap and conduct finite scale analysis of this parameter. Our findings demonstrate that the order parameter effectively captures the amorphous phase transition process across various systems. These results hold significant scientific implications for the study of amorphous structural characteristics via deep learning.

Integrating low-cost sensor monitoring, satellite mapping, and geospatial artificial intelligence for intra-urban air pollution predictions

There is a growing need to apply geospatial artificial intelligence analysis to disparate environmental datasets to find solutions that benefit frontline communities. One such critically needed solution is the prediction of health-relevant ambient ground-level air pollution concentrations. However, many challenges exist surrounding the size and representativeness of limited ground reference stations for model development, reconciling multi-source data, and interpretability of deep learning models. This research addresses these challenges by leveraging a strategically deployed, extensive low-cost sensor (LCS) network that was rigorously calibrated through an optimized neural network. A set of raster predictors with varying data quality and spatial scales was retrieved and processed, including gap-filled satellite aerosol optical depth products and airborne LiDAR-derived 3D urban form. We developed a multi-scale, attention-enhanced convolutional neural network model to reconcile the LCS measurements and multi-source predictors for estimating daily PM2.5 concentration at 30-m resolution. This model employs an advanced approach by using the geostatistical kriging method to generate a baseline pollution pattern and a multi-scale residual method to identify both regional patterns and localized events for high-frequency feature retention. We further used permutation tests to quantify the feature importance, which has rarely been done in DL applications in environmental science. Finally, we demonstrated one application of the model by investigating the air pollution inequality issue across and within various urbanization levels at the block group scale. Overall, this research demonstrates the potential of geospatial AI analysis to provide actionable solutions for addressing critical environmental issues.

Development of a Bispectral index score prediction model based on an interpretable deep learning algorithm

BackgroundProper maintenance of hypnosis is crucial for ensuring the safety of patients undergoing surgery. Accordingly, indicators, such as the Bispectral index (BIS), have been developed to monitor hypnotic levels. However, the black-box nature of the algorithm coupled with the hardware makes it challenging to understand the underlying mechanisms of the algorithms and integrate them with other monitoring systems, thereby limiting their use. ObjectiveWe propose an interpretable deep learning model that forecasts BIS values 25 s in advance using 30 s electroencephalogram (EEG) data. Material and methodsThe proposed model utilized EEG data as a predictor, which is then decomposed into amplitude and phase components using fast Fourier Transform. An attention mechanism was applied to interpret the importance of these components in predicting BIS. The predictability of the model was evaluated on both regression and binary classification tasks, where the former involved predicting a continuous BIS value, and the latter involved classifying a dichotomous status at a BIS value of 60. To evaluate the interpretability of the model, we analyzed the attention values expressed in the amplitude and phase components according to five ranges of BIS values. The proposed model was trained and evaluated using datasets collected from two separate medical institutions. Results and conclusionThe proposed model achieved excellent performance on both the internal and external validation datasets. The model achieved a root-mean-square error of 6.614 for the regression task, and an area under the receiver operating characteristic curve of 0.937 for the binary classification task. Interpretability analysis provided insight into the relationship between EEG frequency components and BIS values. Specifically, the attention mechanism revealed that higher BIS values were associated with increased amplitude attention values in high-frequency bands and increased phase attention values in various frequency bands. This finding is expected to facilitate a more profound understanding of the BIS prediction mechanism, thereby contributing to the advancement of anesthesia technologies.

Abstract 4306: Identifying genetic dependencies of replication stress using interpretable artificial intelligence

Abstract Rapid proliferation is a hallmark of tumor cells, associated with sensitivity to chemicals that cause DNA replication stress (RS). Due to sustained proliferative signaling and/or defective DNA repair, cancer cells undergo persistent RS making them strongly dependent on the replication stress response (RSR). A consequence of this dependency is that RS becomes an exploitable therapeutic vulnerability in cancer treatment. Classical RS drugs work predominantly by interfering with DNA replication in dividing cells. Recently, an increasing number of drugs have been designed to specifically target RSR proteins. However, molecular pathways responsible for drug response are incompletely understood. Here we build an interpretable deep-learning model aimed at understanding mechanisms of susceptibility and resistance to replicative stress. Instead of associating genetic alterations with drug responses directly, our approach is to project individual mutations on a map of protein complexes and larger molecular assemblies with prior evidence for involvement in cancer. This approach is prompted and supported by the concept that cancer is a network-based disease arising from the action of hallmark cancer pathways. Through systematic interpretation, the model identifies 37 complexes that integrate rare alterations in hundreds of genes for accurate response prediction. The complexes, which cover roles in transcription, DNA repair, cell-cycle checkpoint, and immunity, are further investigated by directed genetic perturbations, validating 24 for which RS effects are phenocopied by CRISPR guide RNAs. For complexes with poorly characterized functions, further insights are obtained via their profiles of in-silico activation across RS agents. This study creates a library of system-level genetic vulnerabilities governing replication stress, with implications for drug selection and combination. Citation Format: Xiaoyu Zhao, Akshat Singhal, Trey Ideker. Identifying genetic dependencies of replication stress using interpretable artificial intelligence. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4306.

A Study of CNN and Transfer Learning in Medical Imaging: Advantages, Challenges, Future Scope

This paper presents a comprehensive study of Convolutional Neural Networks (CNN) and transfer learning in the context of medical imaging. Medical imaging plays a critical role in the diagnosis and treatment of diseases, and CNN-based models have demonstrated significant improvements in image analysis and classification tasks. Transfer learning, which involves reusing pre-trained CNN models, has also shown promise in addressing challenges related to small datasets and limited computational resources. This paper reviews the advantages of CNN and transfer learning in medical imaging, including improved accuracy, reduced time and resource requirements, and the ability to address class imbalances. It also discusses challenges, such as the need for large and diverse datasets, and the limited interpretability of deep learning models. What factors contribute to the success of these networks? How are they fashioned, exactly? What motivated them to build the structures that they did? Finally, the paper presents current and future research directions and opportunities, including the development of specialized architectures and the exploration of new modalities and applications for medical imaging using CNN and transfer learning techniques. Overall, the paper highlights the significant potential of CNN and transfer learning in the field of medical imaging, while also acknowledging the need for continued research and development to overcome existing challenges and limitations.

Deep Learning in COVID-19 Diagnosis, Prognosis and Treatment Selection

Deep learning is a sub-discipline of artificial intelligence that uses artificial neural networks, a machine learning technique, to extract patterns and make predictions from large datasets. In recent years, it has achieved rapid development and is widely used in numerous disciplines with fruitful results. Learning valuable information from complex, high-dimensional, and heterogeneous biomedical data is a key challenge in transforming healthcare. In this review, we provide an overview of emerging deep-learning techniques, COVID-19 research involving deep learning, and concrete examples of deep-learning methods in COVID-19 diagnosis, prognosis, and treatment management. Deep learning can process medical imaging data, laboratory test results, and other relevant data to diagnose diseases and judge disease progression and prognosis, and even recommend treatment plans and drug-use strategies to accelerate drug development and improve drug quality. Furthermore, it can help governments develop proper prevention and control measures. We also assess the current limitations and challenges of deep learning in therapy precision for COVID-19, including the lack of phenotypically abundant data and the need for more interpretable deep-learning models. Finally, we discuss how current barriers can be overcome to enable future clinical applications of deep learning.

Graph Node Based Interpretability Guided Sample Selection for Active Learning.

While supervised learning techniques have demonstrated state-of-the-art performance in many medical image analysis tasks, the role of sample selection is important. Selecting the most informative samples contributes to the system attaining optimum performance with minimum labeled samples, which translates to fewer expert interventions and cost. Active Learning (AL) methods for informative sample selection are effective in boosting performance of computer aided diagnosis systems when limited labels are available. Conventional approaches to AL have mostly focused on the single label setting where a sample has only one disease label from the set of possible labels. These approaches do not perform optimally in the multi-label setting where a sample can have multiple disease labels (e.g. in chest X-ray images). In this paper we propose a novel sample selection approach based on graph analysis to identify informative samples in a multi-label setting. For every analyzed sample, each class label is denoted as a separate node of a graph. Building on findings from interpretability of deep learning models, edge interactions in this graph characterize similarity between corresponding interpretability saliency map model encodings. We explore different types of graph aggregation to identify informative samples for active learning. We apply our method to public chest X-ray and medical image datasets, and report improved results over state-of-the-art AL techniques in terms of model performance, learning rates, and robustness.

Real-Time, Model-Agnostic and User-Driven Counterfactual Explanations Using Autoencoders

Explainable Artificial Intelligence (XAI) has gained significant attention in recent years due to concerns over the lack of interpretability of Deep Learning models, which hinders their decision-making processes. To address this issue, counterfactual explanations have been proposed to elucidate the reasoning behind a model’s decisions by providing what-if statements as explanations. However, generating counterfactuals traditionally involves solving an optimization problem for each input, making it impractical for real-time feedback. Moreover, counterfactuals must meet specific criteria, including being user-driven, causing minimal changes, and staying within the data distribution. To overcome these challenges, a novel model-agnostic approach called Real-Time Guided Counterfactual Explanations (RTGCEx) is proposed. This approach utilizes autoencoders to generate real-time counterfactual explanations that adhere to these criteria by optimizing a multiobjective loss function. The performance of RTGCEx has been evaluated on two datasets: MNIST and Gearbox, a synthetic time series dataset. The results demonstrate that RTGCEx outperforms traditional methods in terms of speed and efficacy on MNIST, while also effectively identifying and rectifying anomalies in the Gearbox dataset, highlighting its versatility across different scenarios.

Explainable AI (XAI) for Health Insurance Underwriting

This research paper focuses on offering an understanding of the role of Explainable Artificial Intelligence (XAI) when underwriting a health insurance policy. The latter is due to AI systems being applied in risk management and policy-making across the insurance industry, and hence, there is a rising need for explicability of such systems. This work provides a detailed exploration of several XAI methods, the application of the selected approaches within the case of health insurance, and the barriers to achieving a proper level of model interpretability and reliability. We focus on the national and international methods for interpretation, the locally interpretable deep learning models, the metrics for XAI in underwriting context. Furthermore, we present a brief of important regulatory concerns, ethical issues, and recommendations for further research wherein the field is experiencing rapid expansion. Based on our findings, we conclude that XAI provides viable solutions for building trustworthiness in health insurance underwriting AI systems but state limitations with relation to scalability for handling complex health data and to meet strict regulatory concerns. The study therefore presents XAI as a tool that has the potential of transforming underwriting process especially by improving the amount of trust that health insurance markets and consumers place in the AI models used in underwriting.

An Explainable Spatial-Temporal Graphical Convolutional Network to Score Freezing of Gait in Parkinsonian Patients.

Freezing of gait (FOG) is a poorly understood heterogeneous gait disorder seen in patients with parkinsonism which contributes to significant morbidity and social isolation. FOG is currently measured with scales that are typically performed by movement disorders specialists (i.e., MDS-UPDRS), or through patient completed questionnaires (N-FOG-Q) both of which are inadequate in addressing the heterogeneous nature of the disorder and are unsuitable for use in clinical trials The purpose of this study was to devise a method to measure FOG objectively, hence improving our ability to identify it and accurately evaluate new therapies. A major innovation of our study is that it is the first study of its kind that uses the largest sample size (>30 h, N = 57) in order to apply explainable, multi-task deep learning models for quantifying FOG over the course of the medication cycle and at varying levels of parkinsonism severity. We trained interpretable deep learning models with multi-task learning to simultaneously score FOG (cross-validated F1 score 97.6%), identify medication state (OFF vs. ON levodopa; cross-validated F1 score 96.8%), and measure total PD severity (MDS-UPDRS-III score prediction error ≤ 2.7 points) using kinematic data of a well-characterized sample of N = 57 patients during levodopa challenge tests. The proposed model was able to explain how kinematic movements are associated with each FOG severity level that were highly consistent with the features, in which movement disorders specialists are trained to identify as characteristics of freezing. Overall, we demonstrate that deep learning models' capability to capture complex movement patterns in kinematic data can automatically and objectively score FOG with high accuracy. These models have the potential to discover novel kinematic biomarkers for FOG that can be used for hypothesis generation and potentially as clinical trial outcome measures.

Leukocyte deep learning classification assessment using Shapley additive explanations algorithm.

A peripheral blood smear is a basic test for hematological disease diagnosis. This test is performed manually in many places worldwide, which requires both time and qualified staff. Large laboratories are equipped with digital morphology analyzers, some of which are based on deep learning methods. However, it is difficult to explain to scientists how they work. In this paper, we proposed to add an explanatory factor to enhance the interpretability of deep learning models in leukocyte classification. 10 297 single images of leukocytes obtained from peripheral blood smears were included in this study. Pre-trained and fully trained VGG16 and VGG19 models were used to classify the leukocytes, and Shapley Additive Explanations (SHAP) DeepExplainer was applied to visualize the area of cells that were significant for classification. The output images from the DeepExplainer were compared with cellular elements that are essential to laboratory practice. The accuracy of our fully trained models was 99.81% for VGG16 and 99.79% for VGG19. It achieved slightly better results than the partially trained model, which scored 98.67% for VGG16 and 98.33% for VGG19. Their SHAP explanations indicated the significance of cellular structures in microscopic examination. Explanations in the pre-trained models have proved the cell and nucleus contours to be relevant to classification, while explanations in the fully trained models pointed to the cytoplasm area. Despite different SHAP DeepExplainer explanations for fully and partially trained models, this method appears to be helpful for the verification of leukocyte classification in automated peripheral blood smear examination.

Interpretability of deep learning models for crop yield forecasting

Machine learning models for crop yield forecasting often rely on expert-designed features or predictors. The effectiveness and interpretability of these handcrafted features depends on the expertise of the people designing them. Neural networks have the ability to learn features directly from input data and train the feature learning and prediction steps simultaneously. In this paper, we evaluate the performance and interpretability of neural network models for crop yield forecasting using data from the MARS Crop Yield Forecasting System of the European Commission’s Joint Research Centre. The selected neural networks can handle sequential or time series data and include long short-term memory (LSTM) recurrent neural network and 1-dimensional convolutional neural network (1DCNN). Performance was compared with a linear trend model and a Gradient-Boosted Decision Trees (GBDT) model, trained using hand-designed features. Feature importance scores of input variables were computed using feature attribution methods and were analyzed by crop yield modeling and agronomy experts. Results showed that LSTM models perform statistically better than GBDT models for soft wheat in Germany and similar to GBDT models for all other case studies. In addition, LSTM models captured the effect of yield trend, static features (e.g. elevation, soil water holding capacity) and biomass features on crop yield well, but struggled to capture the impact of extreme temperature and moisture conditions. Our work shows the potential of deep learning to automatically learn features and produce reliable crop yield forecasts, and highlights the importance and challenges of involving human stakeholders in assessing model interpretability.

Interpretable deep dictionary learning for sound speed profiles with uncertainties.

Uncertainties abound in sound speed profiles (SSPs) measured/estimated by modern ocean observing systems, which impede the knowledge acquisition and downstream underwater applications. To reduce the SSP uncertainties and draw insights into specific ocean processes, an interpretable deep dictionary learning model is proposed to cater for uncertain SSP processing. In particular, two kinds of SSP uncertainties are considered: measurement errors, which generally exist in the form of Gaussian noises; and the disturbances/anomalies caused by potential ocean dynamics, which occur at some specific depths and durations. To learn the generative patterns of these uncertainties while maintaining the interpretability of the resulting deep model, the adopted scheme first unrolls the classical K-singular value decomposition algorithm into a neural network, and trains this neural network in a supervised learning manner. The training data and model initializations are judiciously designed to incorporate the environmental properties of ocean SSPs. Experimental results demonstrate the superior performance of the proposed method over the classical baseline in mitigating noise corruptions, detecting, and localizing SSP disturbances/anomalies.