We study the tensor-on-tensor regression, where the goal is to connect tensor responses to tensor covariates with a low Tucker rank parameter tensor/matrix without the prior knowledge of its intrinsic rank. We propose the Riemannian gradient descent (RGD) and Riemannian Gauss-Newton (RGN) methods and cope with the challenge of unknown rank by studying the effect of rank over-parameterization. We provide the first convergence guarantee for the general tensor-on-tensor regression by showing that RGD and RGN respectively converge linearly and quadratically to a statistically optimal estimate in both rank correctly-parameterized and over-parameterized settings. Our theory reveals an intriguing phenomenon: Riemannian optimization methods naturally adapt to over-parameterization without modifications to their implementation. We also prove the statistical-computational gap in scalar-on-tensor regression by a direct low-degree polynomial argument. Our theory demonstrates a "blessing of statistical-computational gap" phenomenon: in a wide range of scenarios in tensor-on-tensor regression for tensors of order three or higher, the computationally required sample size matches what is needed by moderate rank over-parameterization when considering computationally feasible estimators, while there are no such benefits in the matrix settings. This shows moderate rank over-parameterization is essentially "cost-free" in terms of sample size in tensor-on-tensor regression of order three or higher. Finally, we conduct simulation studies to show the advantages of our proposed methods and to corroborate our theoretical findings.

Imputation allows datasets to be used with algorithms that cannot handle missing values by themselves. However, missing values may in principle contribute useful information that is lost through imputation. The missing-indicator approach can be used to preserve this information. There are several theoretical considerations why missing-indicators may or may not be beneficial, but there has not been any large-scale practical experiment on real-life datasets to test this question for machine learning predictions. We perform this experiment for three imputation strategies and a range of different classification algorithms, on the basis of twenty real-life datasets. We find that missing-indicators generally increase classification performance, and that nearest neighbour and iterative imputation do not lead to better performance than simple mean/mode imputation. Therefore, we recommend the use of missing-indicators with mean/mode imputation as a safe default, with the caveat that for decision trees, pruning is necessary to prevent overfitting.

This paper introduces pmuGE (phasor measurement unit Generator of Events), one of the first data-driven generative model for power system event data. We have trained this model on thousands of actual events and created a dataset denoted pmuBAGE (the Benchmarking Assortment of Generated PMU Events). The dataset consists of almost 1000 instances of labeled event data to encourage benchmark evaluations on phasor measurement unit (PMU) data analytics. PMU data are challenging to obtain, especially those covering event periods. Nevertheless, power system problems have recently seen phenomenal advancements via data-driven machine learning solutions. A highly accessible standard benchmarking dataset would enable a drastic acceleration of the development of successful machine learning techniques in this field. We propose a novel learning method based on the Event Participation Decomposition of Power System Events, which makes it possible to learn a generative model of PMU data during system anomalies. The model can create highly realistic event data without compromising the differential privacy of the PMUs used to train it. The dataset is available online for any researcher or practitioner to use at the pmuBAGE Github Repository: https://github.com/NanpengYu/pmuBAGE.

Stack autoencoder (SAE), as a representative deep network, has unique and excellent performance in feature learning, and has received extensive attention from researchers. However, existing deep SAEs focus on original samples without considering the hierarchical structural information between samples. To address this limitation, this paper proposes a new SAE model-neighbouring envelope embedded stack autoencoder ensemble (NE_ESAE). Firstly, the neighbouring sample envelope learning mechanism (NSELM) is proposed for preprocessing of input of SAE. NSELM constructs sample pairs by combining neighbouring samples. Besides, the NSELM constructs a multilayer sample spaces by multilayer iterative mean clustering, which considers the similar samples and generates layers of envelope samples with hierarchical structural information. Second, an embedded stack autoencoder (ESAE) is proposed and trained in each layer of sample space to consider the original samples during training and in the network structure, thereby better finding the relationship between original feature samples and deep feature samples. Third, feature reduction and base classifiers are conducted on the layers of envelope samples respectively, and output classification results of every layer of samples. Finally, the classification results of the layers of envelope sample space are fused through the ensemble mechanism. In the experimental section, the proposed algorithm is validated with over ten representative public datasets. The results show that our method significantly has better performance than existing traditional feature learning methods and the representative deep autoencoders.

In this article, we describe Conditioned Localizer and Classifier (CoLoC) which is a novel solution for Sound Event Localization and Detection (SELD). The solution constitutes of two stages: the localization is done first and is followed by classification conditioned by the output of the localizer. In order to resolve the problem of the unknown number of sources we incorporate the idea borrowed from Sequential Set Generation (SSG). Models from both stages are SELDnet-like CRNNs, but with single outputs. Conducted reasoning shows that such two single-output models are fit for SELD task. We show that our solution improves on the baseline system in most metrics on the STARSS22 Dataset.

Archetypal analysis is an unsupervised machine learning method that summarizes data using a convex polytope. In its original formulation, for fixed k, the method finds a convex polytope with k vertices, called archetype points, such that the polytope is contained in the convex hull of the data and the mean squared Euclidean distance between the data and the polytope is minimal. In the present work, we consider an alternative formulation of archetypal analysis based on the Wasserstein metric, which we call Wasserstein archetypal analysis (WAA). In one dimension, there exists a unique solution of WAA [7] and, in two dimensions, we prove existence of a solution, as long as the data distribution is absolutely continuous with respect to Lebesgue measure. We discuss obstacles to extending our result to higher dimensions and general data distributions. We then introduce an appropriate regularization of the problem, via a Rényi entropy, which allows us to obtain existence of solutions of the regularized problem for general data distributions, in arbitrary dimensions. We prove a consistency result for the regularized problem, ensuring that if the data are iid samples from a probability measure, then as the number of samples is increased, a subsequence of the archetype points converges to the archetype points for the limiting data distribution, almost surely. Finally, we develop and implement a gradient-based computational approach for the twodimensional problem, based on the semi-discrete formulation of the Wasserstein metric. Our analysis is supported by detailed computational experiments.

With the recent surge of NLP technologies in the financial domain, banks and other financial entities have adopted virtual agents (VA) to assist customers. A challenging problem for VAs in this domain is determining a user's reason or intent for contacting the VA, especially when the intent was unseen or open during the VA's training. One method for handling open intents is adaptive decision boundary (ADB) post-processing, which learns tight decision boundaries from intent representations to separate known and open intents. We propose incorporating two methods for supervised pre-training of intent representations: prefix-tuning and fine-tuning just the last layer of a large language model (LLM). With this proposal, our accuracy is 1.63% - 2.07% higher than the prior state-of-the-art ADB method for open intent classification on the banking77 benchmark amongst others. Notably, we only supplement the original ADB model with 0.1% additional trainable parameters. Ablation studies also determine that our method yields better results than full fine-tuning the entire model. We hypothesize that our findings could stimulate a new optimal method of downstream tuning that combines parameter efficient tuning modules with fine-tuning a subset of the base model's layers.

This work demonstrates that Legal Judgement Prediction systems without expert-informed adjustments can be vulnerable to shallow, distracting surface signals that arise from corpus construction, case distribution, and confounding factors. To mitigate this, we use domain expertise to strategically identify statistically predictive but legally irrelevant information. We adopt adversarial training to prevent the system from relying on it. We evaluate our deconfounded models by employing interpretability techniques and comparing to expert annotations. Quantitative experiments and qualitative analysis show that our deconfounded model consistently aligns better with expert rationales than baselines trained for prediction only. We further contribute a set of reference expert annotations to the validation and testing partitions of an existing benchmark dataset of European Court of Human Rights cases.

Contrastive learning is a method of learning visual representations by training Deep Neural Networks (DNNs) to increase the similarity between representations of positive pairs (transformations of the same image) and reduce the similarity between representations of negative pairs (transformations of different images). Here we explore Energy-Based Contrastive Learning (EBCLR) that leverages the power of generative learning by combining contrastive learning with Energy-Based Models (EBMs). EBCLR can be theoretically interpreted as learning the joint distribution of positive pairs, and it shows promising results on small and medium-scale datasets such as MNIST, Fashion-MNIST, CIFAR-10, and CIFAR-100. Specifically, we find EBCLR demonstrates from X4 up to X20 acceleration compared to SimCLR and MoCo v2 in terms of training epochs. Furthermore, in contrast to SimCLR, we observe EBCLR achieves nearly the same performance with 254 negative pairs (batch size 128) and 30 negative pairs (batch size 16) per positive pair, demonstrating the robustness of EBCLR to small numbers of negative pairs. Hence, EBCLR provides a novel avenue for improving contrastive learning methods that usually require large datasets with a significant number of negative pairs per iteration to achieve reasonable performance on downstream tasks. Code: https://github.com/1202kbs/EBCLR

Graph neural networks have achieved state-of-the-art accuracy for graph node classification. However, GNNs are difficult to scale to large graphs, for example frequently encountering out-of-memory errors on even moderate size graphs. Recent works have sought to address this problem using a two-stage approach, which first aggregates data along graph edges, then trains a classifier without using additional graph information. These methods can run on much larger graphs and are orders of magnitude faster than GNNs, but achieve lower classification accuracy. We propose a novel two-stage algorithm based on a simple but effective observation: we should first train a classifier then aggregate, rather than the other way around. We show our algorithm is faster and can handle larger graphs than existing two-stage algorithms, while achieving comparable or higher accuracy than popular GNNs. We also present a theoretical basis to explain our algorithm's improved accuracy, by giving a synthetic nonlinear dataset in which performing aggregation before classification actually decreases accuracy compared to doing classification alone, while our classify then aggregate approach substantially improves accuracy compared to classification alone.