As machine learning (ML) models gain traction in clinical applications, understanding the impact of clinician and societal biases on ML models is increasingly important. While biases can arise in the labels used for model training, the many sources from which these biases arise are not yet well-studied. In this paper, we highlight disparate censorship (i.e., differences in testing rates across patient groups) as a source of label bias that clinical ML models may amplify, potentially causing harm. Many patient risk-stratification models are trained using the results of clinician-ordered diagnostic and laboratory tests of labels. Patients without test results are often assigned a negative label, which assumes that untested patients do not experience the outcome. Since orders are affected by clinical and resource considerations, testing may not be uniform in patient populations, giving rise to disparate censorship. Disparate censorship in patients of equivalent risk leads to undertesting in certain groups, and in turn, more biased labels for such groups. Using such biased labels in standard ML pipelines could contribute to gaps in model performance across patient groups. Here, we theoretically and empirically characterize conditions in which disparate censorship or undertesting affect model performance across subgroups. Our findings call attention to disparate censorship as a source of label bias in clinical ML models.
The electronic design automation (EDA) community has been actively exploring machine learning (ML) for very large-scale integrated computer-aided design (VLSI CAD). Many studies explored learning-based techniques for cross-stage prediction tasks in the design flow to achieve faster design convergence. Although building ML models usually requires a large amount of data, most studies can only generate small internal datasets for validation because of the lack of large public datasets. In this essay, we present the first open-source dataset called CircuitNet for ML tasks in VLSI CAD.
This paper seeks to answer the following question: \textit{"What can we learn by predicting accuracy?"}. Indeed, classification is one of the most popular tasks in machine learning, and many loss functions have been developed to maximize this non-differentiable objective function. Unlike past work on loss function design, which was guided mainly by intuition and theory before being validated by experimentation, here we propose to approach this problem in the opposite way: we seek to extract knowledge by experimentation. This data-driven approach is similar to that used in physics to discover general laws from data. We used a symbolic regression method to automatically find a mathematical expression highly correlated with a linear classifier's accuracy. The formula discovered on more than 260 datasets of embeddings has a Pearson's correlation of 0.96 and a $r^2$ of 0.93. More interestingly, this formula is highly explainable and confirms insights from various previous papers on loss design. We hope this work will open new perspectives in the search for new heuristics leading to a deeper understanding of machine learning theory.
The surge in interest in Artificial Intelligence (AI) over the past decade has been driven almost exclusively by advances in Artificial Neural Networks (ANNs). While ANNs set state-of-the-art performance for many previously intractable problems, the use of global gradient descent necessitates large datasets and computational resources for training, potentially limiting their scalability for real-world domains. Spiking Neural Networks (SNNs) are an alternative to ANNs that use more brain-like artificial neurons and can use local unsupervised learning to rapidly discover sparse recognizable features in the input data. SNNs, however, struggle with dynamical stability and have failed to match the accuracy of ANNs. Here we show how an SNN can overcome many of the shortcomings that have been identified in the literature, including offering a principled solution to the dynamical "vanishing spike problem", to outperform all existing shallow SNNs and equal the performance of an ANN. It accomplishes this while using unsupervised learning with unlabeled data and only 1/50th of the training epochs (labeled data is used only for a simple linear readout layer). This result makes SNNs a viable new method for fast, accurate, efficient, explainable, and re-deployable machine learning with unlabeled data.
Health monitoring applications increasingly rely on machine learning techniques to learn end-user physiological and behavioral patterns in everyday settings. Considering the significant role of wearable devices in monitoring human body parameters, on-device learning can be utilized to build personalized models for behavioral and physiological patterns, and provide data privacy for users at the same time. However, resource constraints on most of these wearable devices prevent the ability to perform online learning on them. To address this issue, it is required to rethink the machine learning models from the algorithmic perspective to be suitable to run on wearable devices. Hyperdimensional computing (HDC) offers a well-suited on-device learning solution for resource-constrained devices and provides support for privacy-preserving personalization. Our HDC-based method offers flexibility, high efficiency, resilience, and performance while enabling on-device personalization and privacy protection. We evaluate the efficacy of our approach using three case studies and show that our system improves the energy efficiency of training by up to $45.8\times$ compared with the state-of-the-art Deep Neural Network (DNN) algorithms while offering a comparable accuracy.
Native language identification (NLI) is the task of training (via supervised machine learning) a classifier that guesses the native language of the author of a text. This task has been extensively researched in the last decade, and the performance of NLI systems has steadily improved over the years. We focus on a different facet of the NLI task, i.e., that of analysing the internals of an NLI classifier trained by an \emph{explainable} machine learning algorithm, in order to obtain explanations of its classification decisions, with the ultimate goal of gaining insight into which linguistic phenomena ``give a speaker's native language away''. We use this perspective in order to tackle both NLI and a (much less researched) companion task, i.e., guessing whether a text has been written by a native or a non-native speaker. Using three datasets of different provenance (two datasets of English learners' essays and a dataset of social media posts), we investigate which kind of linguistic traits (lexical, morphological, syntactic, and statistical) are most effective for solving our two tasks, namely, are most indicative of a speaker's L1. We also present two case studies, one on Spanish and one on Italian learners of English, in which we analyse individual linguistic traits that the classifiers have singled out as most important for spotting these L1s. Overall, our study shows that the use of explainable machine learning can be a valuable tool for th
Fundamental diagrams describe the relationship between speed, flow, and density for some roadway (or set of roadway) configuration(s). These diagrams typically do not reflect, however, information on how speed-flow relationships change as a function of exogenous variables such as curb configuration, weather or other exogenous, contextual information. In this paper we present a machine learning methodology that respects known engineering constraints and physical laws of roadway flux - those that are captured in fundamental diagrams - and show how this can be used to introduce contextual information into the generation of these diagrams. The modeling task is formulated as a probe vehicle trajectory reconstruction problem with Neural Ordinary Differential Equations (Neural ODEs). With the presented methodology, we extend the fundamental diagram to non-idealized roadway segments with potentially obstructed traffic data. For simulated data, we generalize this relationship by introducing contextual information at the learning stage, i.e. vehicle composition, driver behavior, curb zoning configuration, etc, and show how the speed-flow relationship changes as a function of these exogenous factors independent of roadway design.
Recently, Stochastic Gradient Descent (SGD) and its variants have become the dominant methods in the large-scale optimization of machine learning (ML) problems. A variety of strategies have been proposed for tuning the step sizes, ranging from adaptive step sizes to heuristic methods to change the step size in each iteration. Also, momentum has been widely employed in ML tasks to accelerate the training process. Yet, there is a gap in our theoretical understanding of them. In this work, we start to close this gap by providing formal guarantees to a few heuristic optimization methods and proposing improved algorithms. First, we analyze a generalized version of the AdaGrad (Delayed AdaGrad) step sizes in both convex and non-convex settings, showing that these step sizes allow the algorithms to automatically adapt to the level of noise of the stochastic gradients. We show for the first time sufficient conditions for Delayed AdaGrad to achieve almost sure convergence of the gradients to zero. Moreover, we present a high probability analysis for Delayed AdaGrad and its momentum variant in the non-convex setting. Second, we analyze SGD with exponential and cosine step sizes, which are empirically successful but lack theoretical support. We provide the very first convergence guarantees for them in the smooth and non-convex setting, with and without the Polyak-{\L}ojasiewicz (PL) condition. We also show their good property of adaptivity to noise under the PL condition. Third, we study the last iterate of momentum methods. We prove the first lower bound in the convex setting for the last iterate of SGD with constant momentum. Moreover, we investigate a class of Follow-The-Regularized-Leader-based momentum algorithms with increasing momentum and shrinking updates. We show that their last iterate has optimal convergence for unconstrained convex stochastic optimization problems.
DNNs are known to be vulnerable to so-called adversarial attacks that manipulate inputs to cause incorrect results that can be beneficial to an attacker or damaging to the victim. Recent works have proposed approximate computation as a defense mechanism against machine learning attacks. We show that these approaches, while successful for a range of inputs, are insufficient to address stronger, high-confidence adversarial attacks. To address this, we propose DNNSHIELD, a hardware-accelerated defense that adapts the strength of the response to the confidence of the adversarial input. Our approach relies on dynamic and random sparsification of the DNN model to achieve inference approximation efficiently and with fine-grain control over the approximation error. DNNSHIELD uses the output distribution characteristics of sparsified inference compared to a dense reference to detect adversarial inputs. We show an adversarial detection rate of 86% when applied to VGG16 and 88% when applied to ResNet50, which exceeds the detection rate of the state of the art approaches, with a much lower overhead. We demonstrate a software/hardware-accelerated FPGA prototype, which reduces the performance impact of DNNSHIELD relative to software-only CPU and GPU implementations.
Wide use and availability of the machine learning and computer vision techniques allows development of relatively complex monitoring systems in many domains. Besides the traditional industrial domain, new application appears also in biology and agriculture, where we could speak about the detection of infections, parasites and weeds, but also about automated monitoring and early warning systems. This is also connected with the introduction of the easily accessible hardware and development kits such as Arduino, or RaspberryPi family. In this paper, we survey 50 existing papers focusing on the methods of automated beehive monitoring methods using the computer vision techniques, particularly on the pollen and Varroa mite detection together with the bee traffic monitoring. Such systems could also be used for the monitoring of the honeybee colonies and for the inspection of their health state, which could identify potentially dangerous states before the situation is critical, or to better plan periodic bee colony inspections and therefore save significant costs. Later, we also include analysis of the research trends in this application field and we outline the possible direction of the new explorations. Our paper is aimed also at veterinary and apidology professionals and experts, who might not be familiar with machine learning to introduce them to its possibilities, therefore each family of applications is opened by a brief theoretical introduction and motivation related to its base method. We hope that this paper will inspire other scientists to use machine learning techniques for other applications in beehive monitoring.
A datathon is a time-constrained competition involving data science applied to a specific problem. In the past decade, datathons have been shown to be a valuable bridge between fields and expertise . Biomedical data analysis represents a challenging area requiring collaboration between engineers, biologists and physicians to gain a better understanding of patient physiology and of guide decision processes for diagnosis, prognosis and therapeutic interventions to improve care practice. Here, we reflect on the outcomes of an event that we organized in Israel at the end of March 2022 between the MIT Critical Data group, Rambam Health Care Campus (Rambam) and the Technion Israel Institute of Technology (Technion) in Haifa. Participants were asked to complete a survey about their skills and interests, which enabled us to identify current needs in machine learning training for medical problem applications. This work describes opportunities and limitations in medical data science in the Israeli context.
Explanations have gained an increasing level of interest in the AI and Machine Learning (ML) communities in order to improve model transparency and allow users to form a mental model of a trained ML model. However, explanations can go beyond this one way communication as a mechanism to elicit user control, because once users understand, they can then provide feedback. The goal of this paper is to present an overview of research where explanations are combined with interactive capabilities as a mean to learn new models from scratch and to edit and debug existing ones. To this end, we draw a conceptual map of the state-of-the-art, grouping relevant approaches based on their intended purpose and on how they structure the interaction, highlighting similarities and differences between them. We also discuss open research issues and outline possible directions forward, with the hope of spurring further research on this blooming research topic.
The in situ synchrotron high-energy X-ray powder diffraction (XRD) technique is highly utilized by researchers to analyze the crystallographic structures of materials in functional devices (e.g., battery materials) or in complex sample environments (e.g., diamond anvil cells or syntheses reactors). An atomic structure of a material can be identified by its diffraction pattern, along with detailed analysis such as Rietveld refinement which indicates how the measured structure deviates from the ideal structure (e.g., internal stresses or defects). For in situ experiments, a series of XRD images is usually collected on the same sample at different conditions (e.g., adiabatic conditions), yielding different states of matter, or simply collected continuously as a function of time to track the change of a sample over a chemical or physical process. In situ experiments are usually performed with area detectors, collecting 2D images composed of diffraction rings for ideal powders. Depending on the material's form, one may observe different characteristics other than the typical Debye Scherrer rings for a realistic sample and its environments, such as textures or preferred orientations and single crystal diffraction spots in the 2D XRD image. In this work, we present an investigation of machine learning methods for fast and reliable identification and separation of the single crystal diffraction spots in XRD images. The exclusion of artifacts during an XRD image integration process allows a precise analysis of the powder diffraction rings of interest. We observe that the gradient boosting method can consistently produce high accuracy results when it is trained with small subsets of highly diverse datasets. The method dramatically decreases the amount of time spent on identifying and separating single crystal spots in comparison to the conventional method.
Background: As an important branch of machine learning pipelines in medical imaging, radiomics faces two major challenges namely reproducibility and accessibility. In this work, we introduce open-radiomics, a set of radiomics datasets along with a comprehensive radiomics pipeline based on our proposed technical protocol to improve the reproducibility of the results. Methods: We curated large-scale radiomics datasets based on three open-source datasets; BraTS 2020 for high-grade glioma (HGG) versus low-grade glioma (LGG) classification and survival analysis, BraTS 2023 for O6-methylguanine-DNA methyltransferase classification, and non-small cell lung cancer survival analysis from the Cancer Imaging Archive. Using BraTS 2020 Magnetic Resonance Imaging (MRI) dataset, we applied our protocol to 369 brain tumor patients (76 LGG, 293 HGG). Leveraging PyRadiomics for LGG vs. HGG classification, we generated 288 datasets from 4 MRI sequences, 3 binWidths, 6 normalization methods, and 4 tumor subregions. Random Forest classifiers were trained and validated (60%,20%,20%) across 100 different data splits (28,800 test results), evaluating Area Under the Receiver Operating Characteristic Curve (AUROC). Results: Unlike binWidth and image normalization, tumor subregion and imaging sequence significantly affected performance of the models. T1 contrast-enhanced sequence and the union of Necrotic and the non-enhancing tumor core subregions resulted in the highest AUROCs (average test AUROC 0.951, 95% confidence interval of (0.949, 0.952)). Although several settings and data splits (28 out of 28800) yielded test AUROC of 1, they were irreproducible. Conclusion: Our experiments demonstrate the sources of variability in radiomics pipelines (e.g., tumor subregion) can have a significant impact on the results, which may lead to superficial perfect performances that are irreproducible.
The introduction of electronic personal health records (EHR) enables nationwide information exchange and curation among different health care systems. However, the current EHR systems do not provide transparent means for diagnosis support, medical research or can utilize the omnipresent data produced by the personal medical devices. Besides, the EHR systems are centrally orchestrated, which could potentially lead to a single point of failure. Therefore, in this article, we explore novel approaches for decentralizing machine learning over distributed ledgers to create intelligent EHR systems that can utilize information from personal medical devices for improved knowledge extraction. Consequently, we proposed and evaluated a conceptual EHR to enable anonymous predictive analysis across multiple medical institutions. The evaluation results indicate that the decentralized EHR can be deployed over the computing continuum with reduced machine learning time of up to 60% and consensus latency of below 8 seconds.
Reservoir computing is a machine learning approach that can generate a surrogate model of a dynamical system. It can learn the underlying dynamical system using fewer trainable parameters and hence smaller training data sets than competing approaches. Recently, a simpler formulation, known as next-generation reservoir computing, removes many algorithm metaparameters and identifies a well-performing traditional reservoir computer, thus simplifying training even further. Here, we study a particularly challenging problem of learning a dynamical system that has both disparate time scales and multiple co-existing dynamical states (attractors). We compare the next-generation and traditional reservoir computer using metrics quantifying the geometry of the ground-truth and forecasted attractors. For the studied four-dimensional system, the next-generation reservoir computing approach uses $\sim 1.7 \times$ less training data, requires $10^3 \times$ shorter `warm up' time, has fewer metaparameters, and has an $\sim 100\times$ higher accuracy in predicting the co-existing attractor characteristics in comparison to a traditional reservoir computer. Furthermore, we demonstrate that it predicts the basin of attraction with high accuracy. This work lends further support to the superior learning ability of this new machine learning algorithm for dynamical systems.
Internet of Underwater Things (IoUT) have gained rapid momentum over the past decade with applications spanning from environmental monitoring and exploration, defence applications, etc. The traditional IoUT systems use machine learning (ML) approaches which cater the needs of reliability, efficiency and timeliness. However, an extensive review of the various studies conducted highlight the significance of data privacy and security in IoUT frameworks as a predominant factor in achieving desired outcomes in mission critical applications. Federated learning (FL) is a secured, decentralized framework which is a recent development in machine learning, that will help in fulfilling the challenges faced by conventional ML approaches in IoUT. This paper presents an overview of the various applications of FL in IoUT, its challenges, open issues and indicates direction of future research prospects.
Training machine learning models on massive datasets incurs substantial computational costs. To alleviate such costs, there has been a sustained effort to develop data-efficient training methods that can carefully select subsets of the training examples that generalize on par with the full training data. However, existing methods are limited in providing theoretical guarantees for the quality of the models trained on the extracted subsets, and may perform poorly in practice. We propose AdaCore, a method that leverages the geometry of the data to extract subsets of the training examples for efficient machine learning. The key idea behind our method is to dynamically approximate the curvature of the loss function via an exponentially-averaged estimate of the Hessian to select weighted subsets (coresets) that provide a close approximation of the full gradient preconditioned with the Hessian. We prove rigorous guarantees for the convergence of various first and second-order methods applied to the subsets chosen by AdaCore. Our extensive experiments show that AdaCore extracts coresets with higher quality compared to baselines and speeds up training of convex and non-convex machine learning models, such as logistic regression and neural networks, by over 2.9x over the full data and 4.5x over random subsets.
Influenza viruses mutate rapidly and can pose a threat to public health, especially to those in vulnerable groups. Throughout history, influenza A viruses have caused pandemics between different species. It is important to identify the origin of a virus in order to prevent the spread of an outbreak. Recently, there has been increasing interest in using machine learning algorithms to provide fast and accurate predictions for viral sequences. In this study, real testing data sets and a variety of evaluation metrics were used to evaluate machine learning algorithms at different taxonomic levels. As hemagglutinin is the major protein in the immune response, only hemagglutinin sequences were used and represented by position-specific scoring matrix and word embedding. The results suggest that the 5-grams-transformer neural network is the most effective algorithm for predicting viral sequence origins, with approximately 99.54% AUCPR, 98.01% F1 score and 96.60% MCC at a higher classification level, and approximately 94.74% AUCPR, 87.41% F1 score and 80.79% MCC at a lower classification level.
Fair Machine Learning endeavors to prevent unfairness arising in the context of machine learning applications embedded in society. Despite the variety of definitions of fairness and proposed "fair algorithms", there remain unresolved conceptual problems regarding fairness. In this paper, we dissect the role of statistical independence in fairness and randomness notions regularly used in machine learning. Thereby, we are led to a suprising hypothesis: randomness and fairness can be considered equivalent concepts in machine learning. In particular, we obtain a relativized notion of randomness expressed as statistical independence by appealing to Von Mises' century-old foundations for probability. This notion turns out to be "orthogonal" in an abstract sense to the commonly used i.i.d.-randomness. Using standard fairness notions in machine learning, which are defined via statistical independence, we then link the ex ante randomness assumptions about the data to the ex post requirements for fair predictions. This connection proves fruitful: we use it to argue that randomness and fairness are essentially relative and that both concepts should reflect their nature as modeling assumptions in machine learning.