Unforeseen particle accelerator interruptions, also known as interlocks, lead to abrupt operational changes despite being necessary safety measures. These may result in substantial loss of beam time and perhaps even equipment damage. We propose a simple yet powerful binary classification model aiming to forecast such interruptions, in the case of the High Intensity Proton Accelerator complex at the Paul Scherrer Institut. The model is formulated as logistic regression penalized by least absolute shrinkage and selection operator, based on a statistical two sample test to distinguish between unstable and stable states of the accelerator. The primary objective for receiving alarms prior to interlocks is to allow for countermeasures and reduce beam time loss. Hence, a continuous evaluation metric is developed to measure the saved beam time in any period, given the assumption that interlocks could be circumvented by reducing the beam current. The best-performing interlock-to-stable classifier can potentially increase the beam time by around 5 min in a day. Possible instrumentation for fast adjustment of the beam current is also listed and discussed.

Learning representations for graph-structured data is essential for graph analytical tasks. While remarkable progress has been made on static graphs, researches on temporal graphs are still in its beginning stage. The bottleneck of the temporal graph representation learning approach is the neighborhood aggregation strategy, based on which graph attributes share and gather information explicitly. Existing neighborhood aggregation strategies fail to capture either the short-term features or the long-term features of temporal graph attributes, leading to unsatisfactory model performance and even poor robustness and domain generality of the representation learning method. To address this problem, we propose a Frame-level Timeline Modeling (FTM) method that helps to capture both short-term and long-term features and thus learns more informative representations on temporal graphs. In particular, we present a novel link-based framing technique to preserve the short-term features and then incorporate a timeline aggregator module to capture the intrinsic dynamics of graph evolution as long-term features. Our method can be easily assembled with most temporal GNNs. Extensive experiments on common datasets show that our method brings great improvements to the capability, robustness, and domain generality of backbone methods in downstream tasks. Our code can be found at https://github.com/yeeeqichen/FTM.

Cross-validation techniques for risk estimation and model selection are widely used in statistics and machine learning. However, the understanding of the theoretical properties of learning via model selection with cross-validation risk estimation is quite low in face of its widespread use. In this context, this paper presents learning via model selection with cross-validation risk estimation as a general systematic learning framework within classical statistical learning theory and establishes distribution-free deviation bounds in terms of VC dimension, giving detailed proofs of the results and considering both bounded and unbounded loss functions. We also deduce conditions under which the deviation bounds of learning via model selection are tighter than that of learning via empirical risk minimization in the whole hypotheses space, supporting the better performance of model selection frameworks observed empirically in some instances.

The support vector machines (SVM) is a powerful classifier used for binary classification to improve the prediction accuracy. However, the non-differentiability of the SVM hinge loss function can lead to computational difficulties in high dimensional settings. To overcome this problem, we rely on Bernstein polynomial and propose a new smoothed version of the SVM hinge loss called the Bernstein support vector machine (BernSVM), which is suitable for the high dimension $p >> n$ regime. As the BernSVM objective loss function is of the class $C^2$, we propose two efficient algorithms for computing the solution of the penalized BernSVM. The first algorithm is based on coordinate descent with maximization-majorization (MM) principle and the second one is IRLS-type algorithm (iterative re-weighted least squares). Under standard assumptions, we derive a cone condition and a restricted strong convexity to establish an upper bound for the weighted Lasso BernSVM estimator. Using a local linear approximation, we extend the latter result to penalized BernSVM with non convex penalties SCAD and MCP. Our bound holds with high probability and achieves a rate of order $\sqrt{s\log(p)/n}$, where $s$ is the number of active features. Simulation studies are considered to illustrate the prediction accuracy of BernSVM to its competitors and also to compare the performance of the two algorithms in terms of computational timing and error estimation. The use of the proposed method is illustrated through analysis of three large-scale real data examples.

Abstract--Early screening for anxiety and appropriate interventions are essential to reduce the incidence of self-harm and suicide in patients. Due to limited medical resources, traditional methods that overly rely on physician expertise and specialized equipment cannot simultaneously meet the needs for high accuracy and model interpretability . Multimodal data can provide more objective evidence for anxiety screening to improve the accuracy of models. The large amount of noise in multimodal data and the unbalanced nature of the data make the model prone to overfitting. However, it is a non-differentiable problem when high-dimensional and multimodal feature combinations are used as model inputs and incorporated into model training. This causes existing anxiety screening methods based on machine learning and deep learning to be inapplicable. Therefore, we propose a multimodal data-driven anxiety screening framework, namely MMD-AS, and conduct experiments on the collected health data of over 200 seafarers by smartphones. The proposed framework's feature extraction, dimension reduction, feature selection, and anxiety inference are jointly trained to improve the model's performance. In the feature selection step, a feature selection method based on the Improved Fireworks Algorithm is used to solve the non-differentiable problem of feature combination to remove redundant features and search for the ideal feature subset. The experimental results show that our framework outperforms the comparison methods. Furthermore, anxiety disorders are accompanied by immune disorders [2], and interfere with cognitive functions through memory and attention [3], thereby affecting normal life and work. Early anxiety assessment and appropriate interventions can greatly reduce the rate of self-harm and suicide in patients [4]. Psychological scales and routine health checks with professional medical equipment are traditional anxiety screening methods. The Self-rating Anxiety Scale (SAS) [5] and the Generalized Anxiety Disorder-7 (GAD-7) [6] are two psychological scales that are currently used for anxiety screening. Anxiety frequently results in a variety of symptoms or behavioral modifications, such as breathlessness [7], variations in blood pressure [8] and heart rate [9], perspiration, tense muscles, and dizziness [10]. These objective signs can also be used as an important basis for anxiety screening. However, due to the limitation of lacking of medical resources in remote areas and high cost, routine health examinations such as Magnetic Resonance Imaging (MRI) [11], Computed T omography (CT), electrocardiogram (ECG) [12], [13] and electroencephalogram (EEG) [9], [14], may not be available. Haimiao Mo and Shuai Ding are with the School of Management, Hefei University of T echnology, Anhui Hefei 23009, China, also with the Key Laboratory of Process Optimization and Intelligent Decision-Making, Ministry of Education, China.

In many scenarios, the interpretability of machine learning models is a highly required but difficult task. To explain the individual predictions of such models, local model-agnostic approaches have been proposed. However, the process generating the explanations can be, for a user, as mysterious as the prediction to be explained. Furthermore, interpretability methods frequently lack theoretical guarantees, and their behavior on simple models is frequently unknown. While it is difficult, if not impossible, to ensure that an explainer behaves as expected on a cutting-edge model, we can at least ensure that everything works on simple, already interpretable models. In this paper, we present a theoretical analysis of Anchors (Ribeiro et al., 2018): a popular rule-based interpretability method that highlights a small set of words to explain a text classifier's decision. After formalizing its algorithm and providing useful insights, we demonstrate mathematically that Anchors produces meaningful results when used with linear text classifiers on top of a TF-IDF vectorization. We believe that our analysis framework can aid in the development of new explainability methods based on solid theoretical foundations.

We study the problem of training a two-layer neural network (NN) of arbitrary width using stochastic gradient descent (SGD) where the input $\boldsymbol{x}\in \mathbb{R}^d$ is Gaussian and the target $y \in \mathbb{R}$ follows a multiple-index model, i.e., $y=g(\langle\boldsymbol{u_1},\boldsymbol{x}\rangle,...,\langle\boldsymbol{u_k},\boldsymbol{x}\rangle)$ with a noisy link function $g$. We prove that the first-layer weights of the NN converge to the $k$-dimensional principal subspace spanned by the vectors $\boldsymbol{u_1},...,\boldsymbol{u_k}$ of the true model, when online SGD with weight decay is used for training. This phenomenon has several important consequences when $k \ll d$. First, by employing uniform convergence on this smaller subspace, we establish a generalization error bound of $O(\sqrt{{kd}/{T}})$ after $T$ iterations of SGD, which is independent of the width of the NN. We further demonstrate that, SGD-trained ReLU NNs can learn a single-index target of the form $y=f(\langle\boldsymbol{u},\boldsymbol{x}\rangle) + \epsilon$ by recovering the principal direction, with a sample complexity linear in $d$ (up to log factors), where $f$ is a monotonic function with at most polynomial growth, and $\epsilon$ is the noise. This is in contrast to the known $d^{\Omega(p)}$ sample requirement to learn any degree $p$ polynomial in the kernel regime, and it shows that NNs trained with SGD can outperform the neural tangent kernel at initialization. Finally, we also provide compressibility guarantees for NNs using the approximate low-rank structure produced by SGD.

Out-of-distribution (OOD) detection methods assume that they have test ground truths, i.e., whether individual test samples are in-distribution (IND) or OOD. However, in the real world, we do not always have such ground truths, and thus do not know which sample is correctly detected and cannot compute the metric like AUROC to evaluate the performance of different OOD detection methods. In this paper, we are the first to introduce the unsupervised evaluation problem in OOD detection, which aims to evaluate OOD detection methods in real-world changing environments without OOD labels. We propose three methods to compute Gscore as an unsupervised indicator of OOD detection performance. We further introduce a new benchmark Gbench, which has 200 real-world OOD datasets of various label spaces to train and evaluate our method. Through experiments, we find a strong quantitative correlation betwwen Gscore and the OOD detection performance. Extensive experiments demonstrate that our Gscore achieves state-of-the-art performance. Gscore also generalizes well with different IND/OOD datasets, OOD detection methods, backbones and dataset sizes. We further provide interesting analyses of the effects of backbones and IND/OOD datasets on OOD detection performance. The data and code will be available.

The use of machine learning (ML) techniques in the biomedical field has become increasingly important, particularly with the large amounts of data generated by the aftermath of the COVID-19 pandemic. However, due to the complex nature of biomedical datasets and the use of black-box ML models, a lack of trust and adoption by domain experts can arise. In response, interpretable ML (IML) approaches have been developed, but the curse of dimensionality in biomedical datasets can lead to model instability. This paper proposes a novel computational strategy for the stratification of biomedical problem datasets into k-fold cross-validation (CVs) and integrating domain knowledge interpretation techniques embedded into the current state-of-the-art IML frameworks. This approach can improve model stability, establish trust, and provide explanations for outcomes generated by trained IML models. Specifically, the model outcome, such as aggregated feature weight importance, can be linked to further domain knowledge interpretations using techniques like pathway functional enrichment, drug targeting, and repurposing databases. Additionally, involving end-users and clinicians in focus group discussions before and after the choice of IML framework can help guide testable hypotheses, improve performance metrics, and build trustworthy and usable IML solutions in the biomedical field. Overall, this study highlights the potential of combining advanced computational techniques with domain knowledge interpretation to enhance the effectiveness of IML solutions in the context of complex biomedical datasets.

We study the problem of learning a Hamiltonian $H$ to precision $\varepsilon$, supposing we are given copies of its Gibbs state $\rho=\exp(-\beta H)/\operatorname{Tr}(\exp(-\beta H))$ at a known inverse temperature $\beta$. Anshu, Arunachalam, Kuwahara, and Soleimanifar (Nature Physics, 2021, arXiv:2004.07266) recently studied the sample complexity (number of copies of $\rho$ needed) of this problem for geometrically local $N$-qubit Hamiltonians. In the high-temperature (low $\beta$) regime, their algorithm has sample complexity poly$(N, 1/\beta,1/\varepsilon)$ and can be implemented with polynomial, but suboptimal, time complexity. In this paper, we study the same question for a more general class of Hamiltonians. We show how to learn the coefficients of a Hamiltonian to error $\varepsilon$ with sample complexity $S = O(\log N/(\beta\varepsilon)^{2})$ and time complexity linear in the sample size, $O(S N)$. Furthermore, we prove a matching lower bound showing that our algorithm's sample complexity is optimal, and hence our time complexity is also optimal. In the appendix, we show that virtually the same algorithm can be used to learn $H$ from a real-time evolution unitary $e^{-it H}$ in a small $t$ regime with similar sample and time complexity.