Large rotating machines, e.g., compressors, steam turbines, gas turbines, are critical equipment in many process industries such as energy, chemical, and power generation. Due to high rotating speed and tremendous momentum of the rotor, the centrifugal force may lead to flying apart of the rotor parts, which brings a great threat to the operation safety. Early detection and prediction of potential failures could prevent the catastrophic plant downtime and economic loss. In this paper, we divide the operational states of a rotating machine into normal, risky, and high-risk ones based on the time to the moment of failure. Then a cascade classifying algorithm is proposed to predict the states in two steps, first we judge whether the machine is in normal or abnormal condition; for time periods which are predicted as abnormal we further classify them into risky or high-risk states. Moreover, traditional classification model evaluation metrics, such as confusion matrix, true-false accuracy, are static and neglect the online prediction dynamics and uneven wrong-prediction prices. An Online Prediction Ability Index (OPAI) is proposed to select prediction models with consistent online predictions and smaller close-to-downtime prediction errors. Real-world data sets and computational experiments are used to verify the effectiveness of proposed methods.

We show that any memory-constrained, first-order algorithm which minimizes $d$-dimensional, $1$-Lipschitz convex functions over the unit ball to $1/\mathrm{poly}(d)$ accuracy using at most $d^{1.25 - \delta}$ bits of memory must make at least $\tilde{\Omega}(d^{1 + (4/3)\delta})$ first-order queries (for any constant $\delta \in [0, 1/4]$). Consequently, the performance of such memory-constrained algorithms are a polynomial factor worse than the optimal $\tilde{O}(d)$ query bound for this problem obtained by cutting plane methods that use $\tilde{O}(d^2)$ memory. This resolves a COLT 2019 open problem of Woodworth and Srebro.

Recent improvements in representation learning without supervision were driven by self-supervised learning approaches, in particular contrastive learning (CL), which constructs positive and negative samples out of unlabeled dataset via data augmentation (Chen et al., 2020; He et al., 2020; Caron et al., 2020; Ye et al., 2019; Oord et al., 2018; Wu et al., 2018). Subsequent works based on data augmentation also showed promising results for methods based on non-contrastive learning (non-CL), which do not require explicit negative samples (Grill et al., 2020; Richemond et al., 2020; Chen & He, 2021; Zbontar et al., 2021; Tian et al., 2021). However, understanding of how these approaches work, especially of how the learned representations compare--qualitatively and quantitatively--is lagging behind. In this paper, via a combination of empirical and theoretical results, we provide evidence that non-contrastive methods based on data augmentation can lead to substantially worse representations. Most notably, avoiding the collapsed representations has been the key ingredient in prior successes in non-contrastive learning. The collapses was first referred as the complete collapse, that is, all representation vectors shrink into a single point; later a new type of collapses, dimension collapse (Hua et al., 2021; Jing et al., 2022) caught attention as well, that is the embedding vectors only span a lower-dimensional subspace.

Dynamic networks are ubiquitous for modelling sequential graph-structured data, e.g., brain connectome, population flows and messages exchanges. In this work, we consider dynamic networks that are temporal sequences of graph snapshots, and aim at detecting abrupt changes in their structure. This task is often termed network change-point detection and has numerous applications, such as fraud detection or physical motion monitoring. Leveraging a graph neural network model, we design a method to perform online network change-point detection that can adapt to the specific network domain and localise changes with no delay. The main novelty of our method is to use a siamese graph neural network architecture for learning a data-driven graph similarity function, which allows to effectively compare the current graph and its recent history. Importantly, our method does not require prior knowledge on the network generative distribution and is agnostic to the type of change-points; moreover, it can be applied to a large variety of networks, that include for instance edge weights and node attributes. We show on synthetic and real data that our method enjoys a number of benefits: it is able to learn an adequate graph similarity function for performing online network change-point detection in diverse types of change-point settings, and requires a shorter data history to detect changes than most existing state-of-the-art baselines.

If the groups under investigation are pre-specified, i.e., not a function of the observed data, then classical hypothesis tests will control the Type I error rate. However, it is increasingly common to want to test for a difference in means between groups that are defined through the observed data, e.g., via the output of a clustering algorithm. For instance, in single-cell RNA-sequencing analysis, researchers often first cluster the cells, and then test for a difference in the expected gene expression levels between the clusters to quantify up-or down-regulation of genes, annotate known cell types, and identify new cell types (Grün et al., 2015; Aizarani et al., 2019; Lähnemann et al., 2020; Zhang et al., 2019; Doughty & Kerkhoven, 2020). In fact, the inferential challenges resulting from testing data-guided hypotheses have been described as a "grand challenge" in the field of genomics (Lähnemann et al., 2020), and papers in the field continue to overlook this issue: as an example, seurat (Stuart et al., 2019), the state-of-the-art single-cell RNA sequencing analysis tool, tests for differential gene expression between groups obtained via clustering, with a note that "p-values [from these hypotheses] should be interpreted cautiously, as the genes used for clustering are the same genes tested for differential expression." Testing data-guided hypothesis also arises in the field of neuroscience (Kriegeskorte et al., 2009; Button, 2019), social psychology (Hung & Fithian, 2020), and physical sciences (Friederich et al., 2020; Pollice

Individual Treatment Effects (ITE) estimation methods have risen in popularity in the last years. Most of the time, individual effects are better presented as Conditional Average Treatment Effects (CATE). Recently, representation balancing techniques have gained considerable momentum in causal inference from observational data, still limited to continuous (and binary) outcomes. However, in numerous pathologies, the outcome of interest is a (possibly censored) survival time. Our paper proposes theoretical guarantees for a representation balancing framework applied to counterfactual inference in a survival setting using a neural network capable of predicting the factual and counterfactual survival functions (and then the CATE), in the presence of censorship, at the individual level. We also present extensive experiments on synthetic and semisynthetic datasets that show that the proposed extensions outperform baseline methods.

Black-box variational inference (BBVI) now sees widespread use in machine learning and statistics as a fast yet flexible alternative to Markov chain Monte Carlo methods for approximate Bayesian inference. However, stochastic optimization methods for BBVI remain unreliable and require substantial expertise and hand-tuning to apply effectively. In this paper, we propose Robust, Automated, and Accurate BBVI (RAABBVI), a framework for reliable BBVI optimization. RAABBVI is based on rigorously justified automation techniques, includes just a small number of intuitive tuning parameters, and detects inaccurate estimates of the optimal variational approximation. RAABBVI adaptively decreases the learning rate by detecting convergence of the fixed--learning-rate iterates, then estimates the symmetrized Kullback--Leiber (KL) divergence between the current variational approximation and the optimal one. It also employs a novel optimization termination criterion that enables the user to balance desired accuracy against computational cost by comparing (i) the predicted relative decrease in the symmetrized KL divergence if a smaller learning were used and (ii) the predicted computation required to converge with the smaller learning rate. We validate the robustness and accuracy of RAABBVI through carefully designed simulation studies and on a diverse set of real-world model and data examples.

Among the methods offered by machine learning and artificial intelligence, clustering methods are among the most interesting. These methods belong to the class of unsupervised methods, and as such do not suffer from bias or presuppositions, since they do not seek to learn a known rule, but rather to identify unknown links. Their appeal, therefore, lies in their ability to make sense of data whose volume and/or cardinality exceed the processing capabilities of a human. There are in the field of artificial intelligence, two major classes of methods: the supervised approach and the unsupervised approach. They are distinguished by the form of the problem that is submitted to machine learning.

Lazy learning refers to machine learning processes in which generalization of the training data is delayed until a query is made to the system. This type of learning is also known as Instance-based Learning. Lazy classifiers are very useful when working with large datasets that have a few attributes. Learning systems have computation occurring at two different times: training time and consultation times. Training time is the time before the consultation time.

Crop maps are crucial for agricultural monitoring and food management and can additionally support domain-specific applications, such as setting cold supply chain infrastructure in developing countries. Machine learning (ML) models, combined with freely-available satellite imagery, can be used to produce cost-effective and high spatial-resolution crop maps. However, accessing ground truth data for supervised learning is especially challenging in developing countries due to factors such as smallholding and fragmented geography, which often results in a lack of crop type maps or even reliable cropland maps. Our area of interest for this study lies in Himachal Pradesh, India, where we aim at producing an open-access binary cropland map at 10-meter resolution for the Kullu, Shimla, and Mandi districts. To this end, we developed an ML pipeline that relies on Sentinel-2 satellite images time series. We investigated two pixel-based supervised classifiers, support vector machines (SVM) and random forest (RF), which are used to classify per-pixel time series for binary cropland mapping. The ground truth data used for training, validation and testing was manually annotated from a combination of field survey reference points and visual interpretation of very high resolution (VHR) imagery. We trained and validated the models via spatial cross-validation to account for local spatial autocorrelation and selected the RF model due to overall robustness and lower computational cost. We tested the generalization capability of the chosen model at the pixel level by computing the accuracy, recall, precision, and F1-score on hold-out test sets of each district, achieving an average accuracy for the RF (our best model) of 87%. We used this model to generate a cropland map for three districts of Himachal Pradesh, spanning 14,600 km2, which improves the resolution and quality of existing public maps.