Anomaly detection in multivariate time series has emerged as a crucial challenge in time series research, with significant research implications in various fields such as fraud detection, fault diagnosis, and system state estimation. Reconstruction-based models have shown promising potential in recent years for detecting anomalies in time series data. However, due to the rapid increase in data scale and dimensionality, the issues of noise and Weak Identity Mapping (WIM) during time series reconstruction have become increasingly pronounced. To address this, we introduce a novel Adaptive Dynamic Neighbor Mask (ADNM) mechanism and integrate it with the Transformer and Denoising Diffusion Model, creating a new framework for multivariate time series anomaly detection, named Denoising Diffusion Mask Transformer (DDMT). The ADNM module is introduced to mitigate information leakage between input and output features during data reconstruction, thereby alleviating the problem of WIM during reconstruction. The Denoising Diffusion Transformer (DDT) employs the Transformer as an internal neural network structure for Denoising Diffusion Model. It learns the stepwise generation process of time series data to model the probability distribution of the data, capturing normal data patterns and progressively restoring time series data by removing noise, resulting in a clear recovery of anomalies. To the best of our knowledge, this is the first model that combines Denoising Diffusion Model and the Transformer for multivariate time series anomaly detection. Experimental evaluations were conducted on five publicly available multivariate time series anomaly detection datasets. The results demonstrate that the model effectively identifies anomalies in time series data, achieving state-of-the-art performance in anomaly detection.
In the field of global optimization, many existing algorithms face challenges posed by non-convex target functions and high computational complexity or unavailability of gradient information. These limitations, exacerbated by sensitivity to initial conditions, often lead to suboptimal solutions or failed convergence. This is true even for Metaheuristic algorithms designed to amalgamate different optimization techniques to improve their efficiency and robustness. To address these challenges, we develop a sequence of multidimensional integration-based methods that we show to converge to the global optima under some mild regularity conditions. Our probabilistic approach does not require the use of gradients and is underpinned by a mathematically rigorous convergence framework anchored in the nuanced properties of nascent optima distribution. In order to alleviate the problem of multidimensional integration, we develop a latent slice sampler that enjoys a geometric rate of convergence in generating samples from the nascent optima distribution, which is used to approximate the global optima. The proposed Probabilistic Global Optimizer (ProGO) provides a scalable unified framework to approximate the global optima of any continuous function defined on a domain of arbitrary dimension. Empirical illustrations of ProGO across a variety of popular non-convex test functions (having finite global optima) reveal that the proposed algorithm outperforms, by order of magnitude, many existing state-of-the-art methods, including gradient-based, zeroth-order gradient-free, and some Bayesian Optimization methods, in term regret value and speed of convergence. It is, however, to be noted that our approach may not be suitable for functions that are expensive to compute.
The increasing amount of graph data places requirements on the efficiency and scalability of graph neural networks (GNNs), despite their effectiveness in various graph-related applications. Recently, the emerging graph condensation (GC) sheds light on reducing the computational cost of GNNs from a data perspective. It aims to replace the real large graph with a significantly smaller synthetic graph so that GNNs trained on both graphs exhibit comparable performance. However, our empirical investigation reveals that existing GC methods suffer from poor generalization, i.e., different GNNs trained on the same synthetic graph have obvious performance gaps. What factors hinder the generalization of GC and how can we mitigate it? To answer this question, we commence with a detailed analysis and observe that GNNs will inject spectrum bias into the synthetic graph, resulting in a distribution shift. To tackle this issue, we propose eigenbasis matching for spectrum-free graph condensation, named GCEM, which has two key steps: First, GCEM matches the eigenbasis of the real and synthetic graphs, rather than the graph structure, which eliminates the spectrum bias of GNNs. Subsequently, GCEM leverages the spectrum of the real graph and the synthetic eigenbasis to construct the synthetic graph, thereby preserving the essential structural information. We theoretically demonstrate that the synthetic graph generated by GCEM maintains the spectral similarity, i.e., total variation, of the real graph. Extensive experiments conducted on five graph datasets verify that GCEM not only achieves state-of-the-art performance over baselines but also significantly narrows the performance gaps between different GNNs.
Unlike primates, training artificial neural networks on changing data distributions leads to a rapid decrease in performance on old tasks. This phenomenon is commonly referred to as catastrophic forgetting. In this paper, we investigate the representational changes that underlie this performance decrease and identify three distinct processes that together account for the phenomenon. The largest component is a misalignment between hidden representations and readout layers. Misalignment occurs due to learning on additional tasks and causes internal representations to shift. Representational geometry is partially conserved under this misalignment and only a small part of the information is irrecoverably lost. All types of representational changes scale with the dimensionality of hidden representations. These insights have implications for deep learning applications that need to be continuously updated, but may also aid aligning ANN models to the rather robust biological vision.
We propose a general method for semantic representation of images and other data using progressive coding. Semantic coding allows for specific pieces of information to be selectively encoded into a set of measurements that can be highly compressed compared to the size of the original raw data. We consider a hierarchical method of coding where a partial amount of semantic information is first encoded a into a coarse representation of the data, which is then refined by additional encodings that add additional semantic information. Such hierarchical coding is especially well-suited for semantic communication i.e. transferring semantic information over noisy channels. Our proposed method can be considered as a generalization of both progressive image compression and source coding for semantic communication. We present results from experiments on the MNIST and CIFAR-10 datasets that show that progressive semantic coding can provide timely previews of semantic information with a small number of initial measurements while achieving overall accuracy and efficiency comparable to non-progressive methods.
Large language models (LLMs) have exhibited remarkable capabilities in NLP-related tasks such as translation, summarizing, and generation. The application of LLMs in specific areas, notably AIOps (Artificial Intelligence for IT Operations), holds great potential due to their advanced abilities in information summarizing, report analyzing, and ability of API calling. Nevertheless, the performance of current LLMs in AIOps tasks is yet to be determined. Furthermore, a comprehensive benchmark is required to steer the optimization of LLMs tailored for AIOps. Compared with existing benchmarks that focus on evaluating specific fields like network configuration, in this paper, we present \textbf{OpsEval}, a comprehensive task-oriented AIOps benchmark designed for LLMs. For the first time, OpsEval assesses LLMs' proficiency in three crucial scenarios (Wired Network Operation, 5G Communication Operation, and Database Operation) at various ability levels (knowledge recall, analytical thinking, and practical application). The benchmark includes 7,200 questions in both multiple-choice and question-answer (QA) formats, available in English and Chinese. With quantitative and qualitative results, we show how various LLM tricks can affect the performance of AIOps, including zero-shot, chain-of-thought, and few-shot in-context learning. We find that GPT4-score is more consistent with experts than widely used Bleu and Rouge, which can be used to replace automatic metrics for large-scale qualitative evaluations.
Machine unlearning, the study of efficiently removing the impact of specific training points on the trained model, has garnered increased attention of late, driven by the need to comply with privacy regulations like the Right to be Forgotten. Although unlearning is particularly relevant for LLMs in light of the copyright issues they raise, achieving precise unlearning is computationally infeasible for very large models. To this end, recent work has proposed several algorithms which approximate the removal of training data without retraining the model. These algorithms crucially rely on access to the model parameters in order to update them, an assumption that may not hold in practice due to computational constraints or when the LLM is accessed via API. In this work, we propose a new class of unlearning methods for LLMs we call ''In-Context Unlearning'', providing inputs in context and without having to update model parameters. To unlearn a particular training instance, we provide the instance alongside a flipped label and additional correctly labelled instances which are prepended as inputs to the LLM at inference time. Our experimental results demonstrate that these contexts effectively remove specific information from the training set while maintaining performance levels that are competitive with (or in some cases exceed) state-of-the-art unlearning methods that require access to the LLM parameters.
Classification models for electroencephalogram (EEG) data show a large decrease in performance when evaluated on unseen test sub jects. We reduce this performance decrease using new regularization techniques during model training. We propose several graphical models to describe an EEG classification task. From each model, we identify statistical relationships that should hold true in an idealized training scenario (with infinite data and a globally-optimal model) but that may not hold in practice. We design regularization penalties to enforce these relationships in two stages. First, we identify suitable proxy quantities (divergences such as Mutual Information and Wasserstein-1) that can be used to measure statistical independence and dependence relationships. Second, we provide algorithms to efficiently estimate these quantities during training using secondary neural network models. We conduct extensive computational experiments using a large benchmark EEG dataset, comparing our proposed techniques with a baseline method that uses an adversarial classifier. We find our proposed methods significantly increase balanced accuracy on test subjects and decrease overfitting. The proposed methods exhibit a larger benefit over a greater range of hyperparameters than the baseline method, with only a small computational cost at training time. These benefits are largest when used for a fixed training period, though there is still a significant benefit for a subset of hyperparameters when our techniques are used in conjunction with early stopping regularization.
Underwater localization is of great importance for marine observation and building positioning, navigation, timing (PNT) systems that could be widely applied in disaster warning, underwater rescues and resources exploration. The uneven distribution of underwater sound velocity poses great challenge for precise underwater positioning. The current soundline correction positioning method mainly aims at scenarios with known target depth. However, for nodes that are non-cooperative nodes or lack of depth information, soundline tracking strategies cannot work well due to nonunique positional solutions. To tackle this issue, we propose an iterative ray tracing 3D underwater localization (IRTUL) method for stratification compensation. To demonstrate the feasibility of fast stratification compensation, we first derive the signal path as a function of glancing angle, and then prove that the signal propagation time and horizontal propagation distance are monotonic functions of the initial grazing angle, so that fast ray tracing can be achieved. Then, we propose an sound velocity profile (SVP) simplification method, which reduces the computational cost of ray tracing. Experimental results show that the IRTUL has the most significant distance correction in the depth direction, and the average accuracy of IRTUL has been improved by about 3 meters compared to localization model with constant sound velocity. Also, the simplified SVP can significantly improve real-time performance with average accuracy loss less than 0.2 m when used for positioning.
Learning from Label Proportions (LLP) is a learning problem where only aggregate level labels are available for groups of instances, called bags, during training, and the aim is to get the best performance at the instance-level on the test data. This setting arises in domains like advertising and medicine due to privacy considerations. We propose a novel algorithmic framework for this problem that iteratively performs two main steps. For the first step (Pseudo Labeling) in every iteration, we define a Gibbs distribution over binary instance labels that incorporates a) covariate information through the constraint that instances with similar covariates should have similar labels and b) the bag level aggregated label. We then use Belief Propagation (BP) to marginalize the Gibbs distribution to obtain pseudo labels. In the second step (Embedding Refinement), we use the pseudo labels to provide supervision for a learner that yields a better embedding. Further, we iterate on the two steps again by using the second step's embeddings as new covariates for the next iteration. In the final iteration, a classifier is trained using the pseudo labels. Our algorithm displays strong gains against several SOTA baselines (up to 15%) for the LLP Binary Classification problem on various dataset types - tabular and Image. We achieve these improvements with minimal computational overhead above standard supervised learning due to Belief Propagation, for large bag sizes, even for a million samples.