Transformer-based and MLP-based methods have emerged as leading approaches in time series forecasting (TSF). However, real-world time series often show different patterns at different scales, and future changes are shaped by the interplay of these overlapping scales, requiring high-capacity models. While Transformer-based methods excel in capturing long-range dependencies, they suffer from high computational complexities and tend to overfit. Conversely, MLP-based methods offer computational efficiency and adeptness in modeling temporal dynamics, but they struggle with capturing temporal patterns with complex scales effectively. Based on the observation of multi-scale entanglement effect in time series, we propose a novel MLP-based Adaptive Multi-Scale Decomposition (AMD) framework for TSF. Our framework decomposes time series into distinct temporal patterns at multiple scales, leveraging the Multi-Scale Decomposable Mixing (MDM) block to dissect and aggregate these patterns. Complemented by the Dual Dependency Interaction (DDI) block and the Adaptive Multi-predictor Synthesis (AMS) block, our approach effectively models both temporal and channel dependencies and utilizes autocorrelation to refine multi-scale data integration. Comprehensive experiments demonstrate our AMD framework not only overcomes the limitations of existing methods but also consistently achieves state-of-the-art performance across various datasets.

Deep learning (e.g., Transformer) has been widely and successfully used in multivariate time series forecasting (MTSF). Unlike existing methods that focus on training models from a single modal of time series input, large language models (LLMs) based MTSF methods with cross-modal text and time series input have recently shown great superiority, especially with limited temporal data. However, current LLM-based MTSF methods usually focus on adapting and fine-tuning LLMs, while neglecting the distribution discrepancy between textual and temporal input tokens, thus leading to sub-optimal performance. To address this issue, we propose a novel Cross-Modal LLM Fine-Tuning (CALF) framework for MTSF by reducing the distribution discrepancy between textual and temporal data, which mainly consists of the temporal target branch with temporal input and the textual source branch with aligned textual input. To reduce the distribution discrepancy, we develop the cross-modal match module to first align cross-modal input distributions. Additionally, to minimize the modality distribution gap in both feature and output spaces, feature regularization loss is developed to align the intermediate features between the two branches for better weight updates, while output consistency loss is introduced to allow the output representations of both branches to correspond effectively. Thanks to the modality alignment, CALF establishes state-of-the-art performance for both long-term and short-term forecasting tasks with low computational complexity, and exhibits favorable few-shot and zero-shot abilities similar to that in LLMs.

Time series forecasting methods generally fall into two main categories: Channel Independent (CI) and Channel Dependent (CD) strategies. While CI overlooks important covariate relationships, CD captures all dependencies without distinction, introducing noise and reducing generalization. Recent advances in Channel Clustering (CC) aim to refine dependency modeling by grouping channels with similar characteristics and applying tailored modeling techniques. However, coarse-grained clustering struggles to capture complex, time-varying interactions effectively. To address these challenges, we propose TimeFilter, a GNN-based framework for adaptive and fine-grained dependency modeling. After constructing the graph from the input sequence, TimeFilter refines the learned spatial-temporal dependencies by filtering out irrelevant correlations while preserving the most critical ones in a patch-specific manner. Extensive experiments on 13 real-world datasets from diverse application domains demonstrate the state-of-the-art performance of TimeFilter. The code is available at https://github.com/TROUBADOUR000/TimeFilter.

Non-stationarity poses significant challenges for multivariate time series forecasting due to the inherent short-term fluctuations and long-term trends that can lead to spurious regressions or obscure essential long-term relationships. Most existing methods either eliminate or retain non-stationarity without adequately addressing its distinct impacts on short-term and long-term modeling. Eliminating non-stationarity is essential for avoiding spurious regressions and capturing local dependencies in short-term modeling, while preserving it is crucial for revealing long-term cointegration across variates. In this paper, we propose TimeBridge, a novel framework designed to bridge the gap between non-stationarity and dependency modeling in long-term time series forecasting. By segmenting input series into smaller patches, TimeBridge applies Integrated Attention to mitigate short-term non-stationarity and capture stable dependencies within each variate, while Cointegrated Attention preserves non-stationarity to model long-term cointegration across variates. Extensive experiments show that TimeBridge consistently achieves state-of-the-art performance in both short-term and long-term forecasting. Additionally, TimeBridge demonstrates exceptional performance in financial forecasting on the CSI 500 and S&P 500 indices, further validating its robustness and effectiveness. Code is available at https://github.com/Hank0626/TimeBridge.

Low-rank regularization (LRR) has been widely applied in various machine learning tasks, but the associated optimization is challenging. Directly optimizing the rank function under constraints is NP-hard in general. To overcome this difficulty, various relaxations of the rank function were studied. However, optimization of these relaxed LRRs typically depends on singular value decomposition, which is a time-consuming and nondifferentiable operator that cannot be optimized with gradient-based techniques. To address these challenges, in this paper we propose an efficient differentiable approximation of the generalized LRR. The considered LRR form subsumes many popular choices like the nuclear norm, the Schatten-p norm, and various nonconvex relaxations. Our method enables LRR terms to be appended to loss functions in a plug-and-play fashion, and the GPU-friendly operations enable efficient and convenient implementation. Furthermore, convergence analysis is presented, which rigorously shows that both the bias and the variance of our rank estimator rapidly reduce with increased sample size and iteration steps. In the experimental study, the proposed method is applied to various tasks, which demonstrates its versatility and efficiency. Code is available at https://github.com/naiqili/EDLRR.

Autonomous navigation is a fundamental task for robot vacuum cleaners in indoor environments. Since their core function is to clean entire areas, robots inevitably encounter dead zones in cluttered and narrow scenarios. Existing planning methods often fail to escape due to complex environmental constraints, high-dimensional search spaces, and high difficulty maneuvers. To address these challenges, this paper proposes an embodied escaping model that leverages a reinforcement learning-based policy with an efficient action mask for dead zone escaping. To alleviate the issue of the sparse reward in training, we introduce a hybrid training policy that improves learning efficiency. In handling redundant and ineffective action options, we design a novel action representation to reshape the discrete action space with a uniform turning radius. Furthermore, we develop an action mask strategy to select valid actions quickly, balancing precision and efficiency. In real-world experiments, our robot is equipped with a Lidar, IMU, and two-wheel encoders. Extensive quantitative and qualitative experiments across varying difficulty levels demonstrate that our robot can consistently escape from challenging dead zones. Moreover, our approach significantly outperforms compared path planning and reinforcement learning methods in terms of success rate and collision avoidance. A video showcasing our methodology and real-world demonstrations is available at https://youtu.be/kBaaYWGhNuE.

Recent CNN and Transformer-based models tried to utilize frequency and periodicity information for long-term time series forecasting. However, most existing work is based on Fourier transform, which cannot capture fine-grained and local frequency structure. In this paper, we propose a Wavelet-Fourier Transform Network (WFTNet) for long-term time series forecasting. WFTNet utilizes both Fourier and wavelet transforms to extract comprehensive temporal-frequency information from the signal, where Fourier transform captures the global periodic patterns and wavelet transform captures the local ones. Furthermore, we introduce a Periodicity-Weighted Coefficient (PWC) to adaptively balance the importance of global and local frequency patterns. Extensive experiments on various time series datasets show that WFTNet consistently outperforms other state-of-the-art baseline. Code is available at https://github.com/Hank0626/WFTNet.

Convolutional neural network (CNN)-based and Transformer-based methods have recently made significant strides in time series forecasting, which excel at modeling local temporal variations or capturing long-term dependencies. However, real-world time series usually contain intricate temporal patterns, thus making it challenging for existing methods that mainly focus on temporal variations modeling from the 1D time series directly. Based on the intrinsic periodicity of time series, we propose a novel Periodicity Decoupling Framework (PDF) to capture 2D temporal variations of decoupled series for long-term series forecasting. Our PDF mainly consists of three components: multi-periodic decoupling block (MDB), dual variations modeling block (DVMB), and variations aggregation block (VAB). Unlike the previous methods that model 1D temporal variations, our PDF mainly models 2D temporal variations, decoupled from 1D time series by MDB. After that, DVMB attempts to further capture short-term and long-term variations, followed by VAB to make final predictions. Extensive experimental results across seven real-world long-term time series datasets demonstrate the superiority of our method over other state-of-the-art methods, in terms of both forecasting performance and computational efficiency. Code is available at https://github.com/Hank0626/PDF.

Deep neural networks (DNNs) have recently achieved remarkable advancements in time series forecasting (TSF) due to their powerful ability of sequence dependence modeling. To date, existing DNN-based TSF methods still suffer from unreliable predictions for real-world data due to its non-stationarity characteristics, i.e., data distribution varies quickly over time. To mitigate this issue, several normalization methods (e.g., SAN) have recently been specifically designed by normalization in a fixed period/window in the time domain. However, these methods still struggle to capture distribution variations, due to the complex time patterns of time series in the time domain. Based on the fact that wavelet transform can decompose time series into a linear combination of different frequencies, which exhibits distribution variations with time-varying periods, we propose a novel Dual-domain Dynamic Normalization (DDN) to dynamically capture distribution variations in both time and frequency domains. Specifically, our DDN tries to eliminate the non-stationarity of time series via both frequency and time domain normalization in a sliding window way. Besides, our DDN can serve as a plug-in-play module, and thus can be easily incorporated into other forecasting models. Extensive experiments on public benchmark datasets under different forecasting models demonstrate the superiority of our DDN over other normalization methods. Code will be made available following the review process.

Isolation Levels (IL) act as correct contracts between applications and database management systems (DBMSs). The complex code logic and concurrent interactions among transactions make it a hard problem to expose violations of various ILs stated by DBMSs. With the recent proliferation of new DBMSs, especially the cloud ones, there is an urgent demand for a general way to verify various ILs. The core challenges come from the requirements of: (a) lightweight (verifying without modifying the application logic in workloads and the source code of DBMSs), (b) generality (verifying various ILs), and (c) efficiency (performing efficient verification on a long running workload). For lightweight, we propose to deduce transaction dependencies based on time intervals of operations collected from client-sides without touching the source code of DBMSs. For generality, based on a thorough analysis of existing concurrency control protocols, we summarize and abstract four mechanisms which can implement ILs in all commercial DBMSs we have investigated. For efficiency, we design a two-level pipeline to organize and sort massive time intervals in a time and memory conservative way; we propose a mechanism-mirrored verification to simulate the concurrency control protocols implemented in DBMSs for high throughputs. Leopard outperforms existing methods by up to 114× in verification time with a relative small memory usage. In practice, Leopard has a superpower to verify various ILs on any workload running on all commercial DBMSs. Moreover, it has successfully discovered 23 bugs that cannot be found by other existing methods.

A central challenge in password security is to characterize the attacker's guessing curve i.e., what is the probability that the attacker will crack a random user's password within the first G guesses. A key challenge is that the guessing curve depends on the attacker's guessing strategy and the distribution of user passwords both of which are unknown to us. In this work we aim to follow Kerckhoffs's principal and analyze the performance of an optimal attacker who knows the password distribution. Let λGdenote the probability that such an attacker can crack a random user's password within G guesses. We develop several statistically rigorous techniques to upper and lower bound λGgiven N independent samples from the unknown password distribution ${\mathcal{P}}$. We show that our upper/lower bounds on λGhold with high confidence and we apply our techniques to analyze eight large password datasets. Our empirical analysis shows that even state-of-the-art password cracking models are often significantly less guess efficient than an attacker who can optimize its attack based on its (partial) knowledge of the password distribution. We also apply our statistical tools to re-examine different models of the password distribution i.e., the empirical password distribution and Zipf's Law. We find that the empirical distribution closely matches our upper/lower bounds on λGwhen the guessing number G is not too large i.e., G ≪ N. However, for larger values of G our empirical analysis rigorously demonstrates that the empirical distribution (resp. Zipf's Law) overestimates the attacker's success rate. We apply our statistical techniques to upper/lower bound the effectiveness of password throttling mechanisms (key-stretching) which are used to reduce the number of attacker guesses G. Finally, if we are willing to make an additional assumption about the way users respond to password restrictions, we can use our statistical techniques to evaluate the effectiveness of various password composition policies which restrict the passwords that users may select.

In password security a defender would like to identify and warn users with weak passwords. Similarly, the defender may also want to predict what fraction of passwords would be cracked within B guesses as the attacker’s guessing budget B varies from small (online attacker) to large (offline attacker). Towards each of these goals the defender would like to quickly estimate the guessing number for each user password pwd assuming that the attacker uses a password cracking model M i.e., how many password guesses will the attacker check before s/he cracks each user password pwd. Since naïve brute-force enumeration can be prohibitively expensive when the guessing number is very large, Dell’Amico and Filippone [1] developed an efficient Monte Carlo algorithm to estimate the guessing number of a given password pwd. While Dell’Amico and Filippone proved that their estimator is unbiased there is no guarantee that the Monte Carlo estimates are accurate nor does the method provide confidence ranges on the estimated guessing number or even indicate if/when there is a higher degree of uncertainty.Our contributions are as follows: First, we identify theoretical examples where, with high probability, Monte Carlo Strength estimation produces highly inaccurate estimates of individual guessing numbers as well as the entire guessing curve. Second, we introduce Confident Monte Carlo Strength Estimation as an extension of Dell’Amico and Filippone [1]. Given a password our estimator generates an upper and lower bound with the guarantee that, except with probability δ, the true guessing number lies within the given confidence range. Our techniques can also be used to characterize the attacker’s guessing curve. In particular, given a probabilistic password cracking model M we can generate high confidence upper and lower bounds on the fraction of passwords that the attacker will crack as the guessing budget B varies.

Effectively extracting features from a discrete point cloud is necessary for three-dimensional (3D) vehicle tracking. However, a point cloud data set is large and sparsely distributed, hindering the success of vehicle-tracking algorithms. To solve this problem, this paper proposes a 3D vehicle object tracking algorithm based on bounding box similarity measurement. The algorithm includes state prediction, temporal association, trajectory management, state update, and other processes. Also incorporated is a vehicle object temporal association method based on a siamese encoder. The bounding box is encoded into a high-dimensional space, the feature distance is calculated as the time series association cost, and triplet loss was introduced to urge the encoder to learn the geometric similarity of the truth matching box. A 3D Kalman filter and greedy matching are used to effectuate 3D vehicle object tracking algorithm. The experimental results using a KITTI multi-object tracking dataset show that the proposed algorithm can achieve good vehicle tracking performance. In the case of frame loss of point cloud in the frequency reduction simulation, compared with the benchmark method AB3DMOT, the average multi-object tracking accuracy (AMOTA) and the average multi-object tracking accuracy (AMOTP) of the improved method are increased by 2.62% and 1.41% respectively, and the performance is better in the frame loss scene.

The increasing popularity of smartphones, equipped with GPS, provides new opportunities for location-based service (LBS). Among all kinds of LBSs, targeted advertising based on users' locations takes great advantage of the rich location data to improve the accuracy of advertising and thus potentially increase the sellers' profits. However, location-based advertising (LBA) has raised significant privacy concerns, since the location information used in such kinds of services is private information, which the users may not be willing to expose. In this paper, we present POLA, which is a Privacy-preserving prOtocol for Location-based real-time Advertising. In this protocol, we not only preserve the privacy of the location data, we also take the values of advertisers into consideration which is also regarded as private information. We show the privacy-preserving properties of POLA in details. Furthermore, we have conducted simulations to evaluate the performance of POLA. Evaluation results show that POLA achieves privacy preserving LBA with relatively low overhead.