After performing comparison of the performance of seven different machine learning models on detection depression tasks to show that the choice of features is essential, we compare our methods and results with the published work of other researchers. In the end we summarize optimal practices in order that this useful classification solution can be translated to clinical practice with high accuracy and better acceptance.
The adoption of intelligent systems with Artificial Neural Networks (ANNs) embedded in hardware for real-time applications currently faces a growing demand in fields like the Internet of Things (IoT) and Machine to Machine (M2M). However, the application of ANNs in this type of system poses a significant challenge due to the high computational power required to process its basic operations. This paper aims to show an implementation strategy of a Multilayer Perceptron (MLP) type neural network, in a microcontroller (a low-cost, low-power platform). A modular matrix-based MLP with the full classification process was implemented, and also the backpropagation training in the microcontroller. The testing and validation were performed through Hardware in the Loop (HIL) of the Mean Squared Error (MSE) of the training process, classification result, and the processing time of each implementation module. The results revealed a linear relationship between the values of the hyperparameters and the processing time required for classification, also the processing time concurs with the required time for many applications on the fields mentioned above. These findings show that this implementation strategy and this platform can be applied successfully on real-time applications that require the capabilities of ANNs.
We study the black-box attacks on graph neural networks (GNNs) under a novel and realistic constraint: attackers have access to only a subset of nodes in the network, and they can only attack a small number of them. A node selection step is essential under this setup. We demonstrate that the structural inductive biases of GNN models can be an effective source for this type of attacks. Specifically, by exploiting the connection between the backward propagation of GNNs and random walks, we show that the common gradient-based white-box attacks can be generalized to the black-box setting via the connection between the gradient and an importance score similar to PageRank. In practice, we find attacks based on this importance score indeed increase the classification loss by a large margin, but they fail to significantly increase the mis-classification rate. Our theoretical and empirical analyses suggest that there is a discrepancy between the loss and mis-classification rate, as the latter presents a diminishing-return pattern when the number of attacked nodes increases. Therefore, we propose a greedy procedure to correct the importance score that takes into account of the diminishing-return pattern. Experimental results show that the proposed procedure can significantly increase the mis-classification rate of common GNNs on real-world data without access to model parameters nor predictions.
Motivated by the problem of optimization of force-field systems in physics using large-scale computer simulations, we consider exploration of a deterministic complex multivariate response surface. The objective is to find input combinations that generate output close to some desired or "target" vector. In spite of reducing the problem to exploration of the input space with respect to a one-dimensional loss function, the search is nontrivial and challenging due to infeasible input combinations, high dimensionalities of the input and output space and multiple "desirable" regions in the input space and the difficulty of emulating the objective function well with a surrogate model. We propose an approach that is based on combining machine learning techniques with smart experimental design ideas to locate multiple good regions in the input space.
Objectives: To investigate machine-learning classifiers and interpretable models using chest CT for detection of COVID-19 and differentiation from other pneumonias, ILD and normal CTs. Methods: Our retrospective multi-institutional study obtained 2096 chest CTs from 16 institutions (including 1077 COVID-19 patients). Training/testing cohorts included 927/100 COVID-19, 388/33 ILD, 189/33 other pneumonias, and 559/34 normal (no pathologies) CTs. A metric-based approach for classification of COVID-19 used interpretable features, relying on logistic regression and random forests. A deep learning-based classifier differentiated COVID-19 via 3D features extracted directly from CT attenuation and probability distribution of airspace opacities. Results: Most discriminative features of COVID-19 are percentage of airspace opacity and peripheral and basal predominant opacities, concordant with the typical characterization of COVID-19 in the literature. Unsupervised hierarchical clustering compares feature distribution across COVID-19 and control cohorts. The metrics-based classifier achieved AUC=0.83, sensitivity=0.74, and specificity=0.79 of versus respectively 0.93, 0.90, and 0.83 for the DL-based classifier. Most of ambiguity comes from non-COVID-19 pneumonia with manifestations that overlap with COVID-19, as well as mild COVID-19 cases. Non-COVID-19 classification performance is 91% for ILD, 64% for other pneumonias and 94% for no pathologies, which demonstrates the robustness of our method against different compositions of control groups. Conclusions: Our new method accurately discriminates COVID-19 from other types of pneumonia, ILD, and no pathologies CTs, using quantitative imaging features derived from chest CT, while balancing interpretability of results and classification performance, and therefore may be useful to facilitate diagnosis of COVID-19.
Person Re-identification (ReID) aims at matching a person of interest across images. In convolutional neural network (CNN) based approaches, loss design plays a vital role in pulling closer features of the same identity and pushing far apart features of different identities. In recent years, triplet loss achieves superior performance and is predominant in ReID. However, triplet loss considers only three instances of two classes in per-query optimization (with an anchor sample as query) and it is actually equivalent to a two-class classification. There is a lack of loss design which enables the joint optimization of multiple instances (of multiple classes) within per-query optimization for person ReID. In this paper, we introduce a multi-class classification loss, i.e., N-tuple loss, to jointly consider multiple (N) instances for per-query optimization. This in fact aligns better with the ReID test/inference process, which conducts the ranking/comparisons among multiple instances. Furthermore, for more efficient multi-class classification, we propose a new meta prototypical N-tuple loss. With the multi-class classification incorporated, our model achieves the state-of-the-art performance on the benchmark person ReID datasets.
Deep neural networks and other sophisticated machine learning models are widely applied to biomedical signal data because they can detect complex patterns and compute accurate predictions. However, the difficulty of interpreting such models is a limitation, especially for applications involving high-stakes decision, including the identification of bacterial infections. In this paper, we consider fast Raman spectroscopy data and demonstrate that a logistic regression model with carefully selected features achieves accuracy comparable to that of neural networks, while being much simpler and more transparent. Our analysis leverages wavelet features with intuitive chemical interpretations, and performs controlled variable selection with knockoffs to ensure the predictors are relevant and non-redundant. Although we focus on a particular data set, the proposed approach is broadly applicable to other types of signal data for which interpretability may be important.
Adversarial examples can cause catastrophic mistakes in Deep Neural Network (DNNs) based vision systems e.g., for classification, segmentation and object detection. The vulnerability of DNNs against such attacks can prove a major roadblock towards their real-world deployment. Transferability of adversarial examples demand generalizable defenses that can provide cross-task protection. Adversarial training that enhances robustness by modifying target model's parameters lacks such generalizability. On the other hand, different input processing based defenses fall short in the face of continuously evolving attacks. In this paper, we take the first step to combine the benefits of both approaches and propose a self-supervised adversarial training mechanism in the input space. By design, our defense is a generalizable approach and provides significant robustness against the \textbf{unseen} adversarial attacks (\eg by reducing the success rate of translation-invariant \textbf{ensemble} attack from 82.6\% to 31.9\% in comparison to previous state-of-the-art). It can be deployed as a plug-and-play solution to protect a variety of vision systems, as we demonstrate for the case of classification, segmentation and detection. Code is available at: {\small\url{https://github.com/Muzammal-Naseer/NRP}}.
Receiver operating characteristic (ROC) curve is an informative tool in binary classification and Area Under ROC Curve (AUC) is a popular metric for reporting performance of binary classifiers. In this paper, first we present a comprehensive review of ROC curve and AUC metric. Next, we propose a modified version of AUC that takes confidence of the model into account and at the same time, incorporates AUC into Binary Cross Entropy (BCE) loss used for training a Convolutional neural Network for classification tasks. We demonstrate this on three datasets: MNIST, prostate MRI, and brain MRI. Furthermore, we have published GenuineAI, a new python library, which provides the functions for conventional AUC and the proposed modified AUC along with metrics including sensitivity, specificity, recall, precision, and F1 for each point of the ROC curve.
Anterior chamber angle (ACA) classification is a key step in the diagnosis of angle-closure glaucoma in Anterior Segment Optical Coherence Tomography (AS-OCT). Existing automated analysis methods focus on a binary classification system (i.e., open angle or angle-closure) in a 2D AS-OCT slice. However, clinical diagnosis requires a more discriminating ACA three-class system (i.e., open, narrow, or synechiae angles) for the benefit of clinicians who seek better to understand the progression of the spectrum of angle-closure glaucoma types. To address this, we propose a novel sequence multi-scale aggregation deep network (SMA-Net) for open-narrow-synechiae ACA classification based on an AS-OCT sequence. In our method, a Multi-Scale Discriminative Aggregation (MSDA) block is utilized to learn the multi-scale representations at slice level, while a ConvLSTM is introduced to study the temporal dynamics of these representations at sequence level. Finally, a multi-level loss function is used to combine the slice-based and sequence-based losses. The proposed method is evaluated across two AS-OCT datasets. The experimental results show that the proposed method outperforms existing state-of-the-art methods in applicability, effectiveness, and accuracy. We believe this work to be the first attempt to classify ACAs into open, narrow, or synechia types grading using AS-OCT sequences.
In this paper, we seek solutions for reducing the computation complexity of transformer-based models for speech representation learning. We evaluate 10 attention algorithms; then, we pre-train the transformer-based model with those attention algorithms in a self-supervised fashion and treat them as feature extractors on downstream tasks, including phoneme classification and speaker classification. With the assistance of t-SNE, PCA and some observation, the attention weights in self-supervised audio transformers can be categorized into four general cases. Based on these cases and some analyses, we are able to use a specific set of attention weights to initialize the model. Our approach shows comparable performance to the typical self-attention yet requires 20% less time in both training and inference.
It is well known that machine learning methods can be vulnerable to adversarially-chosen perturbations of their inputs. Despite significant progress in the area, foundational open problems remain. In this paper, we address several key questions. We derive exact and approximate Bayes-optimal robust classifiers for the important setting of two- and three-class Gaussian classification problems with arbitrary imbalance, for $\ell_2$ and $\ell_\infty$ adversaries. In contrast to classical Bayes-optimal classifiers, determining the optimal decisions here cannot be made pointwise and new theoretical approaches are needed. We develop and leverage new tools, including recent breakthroughs from probability theory on robust isoperimetry, which, to our knowledge, have not yet been used in the area. Our results reveal fundamental tradeoffs between standard and robust accuracy that grow when data is imbalanced. We also show further results, including an analysis of classification calibration for convex losses in certain models, and finite sample rates for the robust risk.
Since its appearance, Visual Question Answering (VQA, i.e. answering a question posed over an image), has always been treated as a classification problem over a set of predefined answers. Despite its convenience, this classification approach poorly reflects the semantics of the problem limiting the answering to a choice between independent proposals, without taking into account the similarity between them (e.g. equally penalizing for answering cat or German shepherd instead of dog). We address this issue by proposing (1) two measures of proximity between VQA classes, and (2) a corresponding loss which takes into account the estimated proximity. This significantly improves the generalization of VQA models by reducing their language bias. In particular, we show that our approach is completely model-agnostic since it allows consistent improvements with three different VQA models. Finally, by combining our method with a language bias reduction approach, we report SOTA-level performance on the challenging VQAv2-CP dataset.
In this paper, we explore the slot tagging with only a few labeled support sentences (a.k.a. few-shot). Few-shot slot tagging faces a unique challenge compared to the other few-shot classification problems as it calls for modeling the dependencies between labels. But it is hard to apply previously learned label dependencies to an unseen domain, due to the discrepancy of label sets. To tackle this, we introduce a collapsed dependency transfer mechanism into the conditional random field (CRF) to transfer abstract label dependency patterns as transition scores. In the few-shot setting, the emission score of CRF can be calculated as a word's similarity to the representation of each label. To calculate such similarity, we propose a Label-enhanced Task-Adaptive Projection Network (L-TapNet) based on the state-of-the-art few-shot classification model -- TapNet, by leveraging label name semantics in representing labels. Experimental results show that our model significantly outperforms the strongest few-shot learning baseline by 14.64 F1 scores in the one-shot setting.
Being an emerging class of in-memory computing architecture, brain-inspired hyperdimensional computing (HDC) mimics brain cognition and leverages random hypervectors (i.e., vectors with a dimensionality of thousands or even more) to represent features and to perform classification tasks. The unique hypervector representation enables HDC classifiers to exhibit high energy efficiency, low inference latency and strong robustness against hardware-induced bit errors. Consequently, they have been increasingly recognized as an appealing alternative to or even replacement of traditional deep neural networks (DNNs) for local on device classification, especially on low-power Internet of Things devices. Nonetheless, unlike their DNN counterparts, state-of-the-art designs for HDC classifiers are mostly security-oblivious, casting doubt on their safety and immunity to adversarial inputs. In this paper, we study for the first time adversarial attacks on HDC classifiers and highlight that HDC classifiers can be vulnerable to even minimally-perturbed adversarial samples. Concretely, using handwritten digit classification as an example, we construct a HDC classifier and formulate a grey-box attack problem, where an attacker's goal is to mislead the target HDC classifier to produce erroneous prediction labels while keeping the amount of added perturbation noise as little as possible. Then, we propose a modified genetic algorithm to generate adversarial samples within a reasonably small number of queries. Our results show that adversarial images generated by our algorithm can successfully mislead the HDC classifier to produce wrong prediction labels with a high probability (i.e., 78% when the HDC classifier uses a fixed majority rule for decision). Finally, we also present two defense strategies -- adversarial training and retraining-- to strengthen the security of HDC classifiers.
We introduce a self-supervised approach for learning node and graph level representations by contrasting structural views of graphs. We show that unlike visual representation learning, increasing the number of views to more than two or contrasting multi-scale encodings do not improve performance, and the best performance is achieved by contrasting encodings from first-order neighbors and a graph diffusion. We achieve new state-of-the-art results in self-supervised learning on 8 out of 8 node and graph classification benchmarks under the linear evaluation protocol. For example, on Cora (node) and Reddit-Binary (graph) classification benchmarks, we achieve 86.8% and 84.5% accuracy, which are 5.5% and 2.4% relative improvements over previous state-of-the-art. When compared to supervised baselines, our approach outperforms them in 4 out of 8 benchmarks. Source code is released at: https://github.com/kavehhassani/mvgrl
Radar-based road user detection is an important topic in the context of autonomous driving applications. The resolution of conventional automotive radar sensors results in a sparse data representation which is tough to refine during subsequent signal processing. On the other hand, a new sensor generation is waiting in the wings for its application in this challenging field. In this article, two sensors of different radar generations are evaluated against each other. The evaluation criterion is the performance on moving road user object detection and classification tasks. To this end, two data sets originating from an off-the-shelf radar and a high resolution next generation radar are compared. Special attention is given on how the two data sets are assembled in order to make them comparable. The utilized object detector consists of a clustering algorithm, a feature extraction module, and a recurrent neural network ensemble for classification. For the assessment, all components are evaluated both individually and, for the first time, as a whole. This allows for indicating where overall performance improvements have their origin in the pipeline. Furthermore, the generalization capabilities of both data sets are evaluated and important comparison metrics for automotive radar object detection are discussed. Results show clear benefits of the next generation radar. Interestingly, those benefits do not actually occur due to better performance at the classification stage, but rather because of the vast improvements at the clustering stage.
Objective: Optical coherence tomography (OCT) and its angiography (OCTA) have several advantages for the early detection and diagnosis of diabetic retinopathy (DR). However, automated, complete DR classification frameworks based on both OCT and OCTA data have not been proposed. In this study, a convolutional neural network (CNN) based method is proposed to fulfill a DR classification framework using en face OCT and OCTA. Methods: A densely and continuously connected neural network with adaptive rate dropout (DcardNet) is designed for the DR classification. In addition, adaptive label smoothing was proposed and used to suppress overfitting. Three separate classification levels are generated for each case based on the International Clinical Diabetic Retinopathy scale. At the highest level the network classifies scans as referable or non-referable for DR. The second level classifies the eye as non-DR, non-proliferative DR (NPDR), or proliferative DR (PDR). The last level classifies the case as no DR, mild and moderate NPDR, severe NPDR, and PDR. Results: We used 10-fold cross-validation with 10% of the data to assess the networks performance. The overall classification accuracies of the three levels were 95.7%, 85.0%, and 71.0% respectively. Conclusion/Significance: A reliable, sensitive and specific automated classification framework for referral to an ophthalmologist can be a key technology for reducing vision loss related to DR.
The identification of Dialogue Act’s (DA) is an important aspect in determining the meaning of an utterance for many applications that require natural language understanding, and recent work using recurrent neural networks (RNN) has shown promising results when applied to the DA classification problem. This work presents a novel probabilistic method of utterance representation and describes a RNN sentence model for out-of-context DA Classification. The utterance representations are generated from keywords selected for their frequency association with certain DA’s. The proposed probabilistic representations are applied to the Switchboard DA corpus and performance is compared with pre-trained word embeddings using the same baseline RNN model. The results indicate that the probabilistic method achieves 75.48% overall accuracy and an improvement over the word embedding representations of 1.8%. This demonstrates the potential utility of using statistical utterance representations, that are able to capture word-DA relationships, for the purpose of DA classification.
Wearable devices have strict power and memory limitations. As a result, there is a need to optimize the power consumption on those devices without sacrificing the accuracy. This paper presents AdaSense: a sensing, feature extraction and classification co-optimized framework for Human Activity Recognition. The proposed techniques reduce the power consumption by dynamically switching among different sensor configurations as a function of the user activity. The framework selects configurations that represent the pareto-frontier of the accuracy and energy trade-off. AdaSense also uses low-overhead processing and classification methodologies. The introduced approach achieves 69% reduction in the power consumption of the sensor with less than 1.5% decrease in the activity recognition accuracy.