Time series forecasting is a long-standing and highly challenging research topic. Recently, driven by the rise of large language models (LLMs), research has increasingly shifted from purely time series methods toward harnessing textual modalities to enhance forecasting performance. However, the vast discrepancy between text and temporal data often leads current multimodal architectures to over-emphasise one modality while neglecting the other, resulting in information loss that harms forecasting performance. To address this modality imbalance, we introduce BALM-TSF (Balanced Multimodal Alignment for LLM-Based Time Series Forecasting), a lightweight time series forecasting framework that maintains balance between the two modalities. Specifically, raw time series are processed by the time series encoder, while descriptive statistics of raw time series are fed to an LLM with learnable prompt, producing compact textual embeddings. To ensure balanced cross-modal context alignment of time series and textual embeddings, a simple yet effective scaling strategy combined with a contrastive objective then maps these textual embeddings into the latent space of the time series embeddings. Finally, the aligned textual semantic embeddings and time series embeddings are together integrated for forecasting. Extensive experiments on standard benchmarks show that, with minimal trainable parameters, BALM-TSF achieves state-of-the-art performance in both long-term and few-shot forecasting, confirming its ability to harness complementary information from text and time series. Code is available at https://github.com/ShiqiaoZhou/BALM-TSF.

Future sustainable energy systems require the integration of local renewable energy sources (RES) into decentralized micro-grids, each containing RES, energy storage systems, and local loads. A substantial challenge associated with micro-grids is the optimization of energy flows to minimize operating costs. This is particularly complex due to (a) the fluctuating power generation of RES, (b) the variability of local loads, and (c) the possibility of energy trade between a micro-grid and a larger ‘utility grid’ that it connects to. Existing methods struggle to manage these sources of uncertainty effectively. To address this, we propose MicroPPO, a reinforcement learning approach for real-time management of power flows in such small-scale energy systems. MicroPPO introduces a novel definition of the environment as a Markov Decision Process (MDP) with a continuous and multi-dimensional action space. This enables more precise control of power flows compared to discrete methods. Additionally, MicroPPO employs an innovative actor network architecture featuring multiple network branches to reflect the individual action dimensions. It further integrates a differentiable projection layer that enforces the feasibility of actions. We assess the performance of our approach against state-of-the-art methods using real-world data. Our results demonstrate MicroPPO's superior convergence towards near-optimal policies.

We study the stochastic Budgeted Multi-Armed Bandit (MAB) problem, where a player chooses from K arms with unknown expected rewards and costs. The goal is to maximize the total reward under a budget constraint. A player thus seeks to choose the arm with the highest reward-cost ratio as often as possible. Current approaches for this problem have several issues, which we illustrate. To overcome them, we propose a new upper confidence bound (UCB) sampling policy, ømega-UCB, that uses asymmetric confidence intervals. These intervals scale with the distance between the sample mean and the bounds of a random variable, yielding a more accurate and tight estimation of the reward-cost ratio compared to our competitors. We show that our approach has sublinear instance-dependent regret in general and logarithmic regret for parameter ρ ≥ 1, and that it outperforms existing policies consistently in synthetic and real settings.

Abstract Change detection is of fundamental importance when analyzing data streams. Detecting changes both quickly and accurately enables monitoring and prediction systems to react, e.g., by issuing an alarm or by updating a learning algorithm. However, detecting changes is challenging when observations are high-dimensional. In high-dimensional data, change detectors should not only be able to identify when changes happen, but also in which subspace they occur. Ideally, one should also quantify how severe they are. Our approach, ABCD, has these properties. ABCD learns an encoder-decoder model and monitors its accuracy over a window of adaptive size. ABCD derives a change score based on Bernstein’s inequality to detect deviations in terms of accuracy, which indicate changes. Our experiments demonstrate that ABCD outperforms its best competitor by up to 20% in F1-score on average. It can also accurately estimate changes’ subspace, together with a severity measure that correlates with the ground truth.

Today, the collection of decentralized data is a common scenario: smartphones store users’ messages locally, smart meters collect energy consumption data, and modern power tools monitor operator behavior. We identify different types of outliers in such data: local, global, and partition outliers. They contain valuable information, for example, about mistakes in operation. However, existing outlier detection approaches cannot distinguish between those types. Thus, we propose a “tandem” technique to join “local” and “federated” outlier detectors. Our core idea is to combine outlier detection on a single device with latent information about devices’ data to discriminate between different outlier types. To the best of our knowledge, our method is the first to achieve this. We evaluate our approach on publicly available synthetic and real-world data that we collect in a study with 15 participants operating power tools.

Abstract Estimating dependencies from data is a fundamental task of Knowledge Discovery. Identifying the relevant variables leads to a better understanding of data and improves both the runtime and the outcomes of downstream Data Mining tasks. Dependency estimation from static numerical data has received much attention. However, real-world data often occurs as heterogeneous data streams: On the one hand, data is collected online and is virtually infinite. On the other hand, the various components of a stream may be of different types, e.g., numerical, ordinal or categorical. For this setting, we propose Monte Carlo Dependency Estimation (MCDE), a framework that quantifies multivariate dependency as the average statistical discrepancy between marginal and conditional distributions, via Monte Carlo simulations. MCDE handles heterogeneity by leveraging three statistical tests: the Mann–Whitney U, the Kolmogorov–Smirnov and the Chi-Squared test. We demonstrate that MCDE goes beyond the state of the art regarding dependency estimation by meeting a broad set of requirements. Finally, we show with a real-world use case that MCDE can discover useful patterns in heterogeneous data streams.

Nowadays, it is common to classify collections of documents into (human-generated, domain-specific) directory structures, such as email or document folders. But documents may be classified wrongly, for a multitude of reasons. Then they are outlying w.r.t. the folder they end up in. Orthogonally to this, and more specifically, two kinds of errors can occur: (O) Out-of-distribution: the document does not belong to any existing folder in the directory; and (M) Misclassification: the document belongs to another folder. It is this specific combination of issues that we address in this article, i.e., we mine text outliers from massive document directories, considering both error types. We propose a new proximity-based algorithm, which we dub kj-Nearest Neighbours (kj-NN). Our algorithm detects text outliers by exploiting semantic similarities and introduces a self-supervision mechanism that estimates the relevance of the original labels. Our approach is efficient and robust to large proportions of outliers. kj-NN also promotes the interpretability of the results by proposing alternative label names and by finding the most similar documents for each outlier. Our real-world experiments demonstrate that our approach outperforms the competitors by a large margin.

The Multi-Armed Bandit (MAB) is a fundamental model capturing the dilemma between exploration and exploitation in sequential decision making. At every time step, the decision maker selects a set of arms and observes a reward from each of the chosen arms. In this paper, we present a variant of the problem, which we call the Scaling MAB (S-MAB): The goal of the decision maker is not only to maximize the cumulative rewards, i.e., choosing the arms with the highest expected reward, but also to decide how many arms to select so that, in expectation, the cost of selecting arms does not exceed the rewards. This problem is relevant to many real-world applications, e.g., online advertising, financial investments or data stream monitoring. We propose an extension of Thompson Sampling, which has strong theoretical guarantees and is reported to perform well in practice. Our extension dynamically controls the number of arms to draw. Furthermore, we combine the proposed method with ADWIN, a state-of-the-art change detector, to deal with non-static environments. We illustrate the benefits of our contribution via a real-world use case on predictive maintenance.

Estimating dependency is a fundamental task in data management. Identifying the relevant variables leads to better understanding and improves both the runtime and outcome of data analysis. In this paper, we propose Monte Carlo Dependency Estimation (MCDE), a framework to estimate multivariate dependency. MCDE quantifies dependency as the average discrepancy between marginal and conditional distributions via Monte Carlo simulations. Based on this framework, we present Mann-Whitney P (MWP), a novel dependency estimator. We show that MWP satisfies a number of desirable properties and demonstrate the superiority of our estimator against the state-of-the-art multivariate dependency measures.

The increasing size of the available data and database volumes represents a real challenge for the data management community. In general, current approaches in data mining require the data to be first extracted from an underlying database. From a practical point of view, this presents many drawbacks. In this short article, we present a possible solution to bridge the gap between data repositories and end user analysis. We demonstrate the interestingness of this approach with ibmdbpy, an open source Python interface developed by IBM for database administration and data analytics.