Buildings energy efficiency is a widely researched topic, which is rapidly gaining popularity due to rising environmental concerns and the need for energy independence. In Northern Europe heating energy alone accounts for up to 70 percent of the total building energy consumption. Industry 4.0 technologies such as IoT, big data, cloud computing and machine learning, along with the creation of predictive and proactive digital twins, can help to reduce this number. However, buildings thermal dynamics is a very complex process that depends on many variables. As a result, commonly used physics-based white box models are time-consuming and require vast expertise. On the contrary, black box forecasting models, which rely primarily on building energy consumption data, lack fundamental insights and hinder re-use. In this study we propose an architecture to facilitate grey box modelling of building thermal dynamics while integrating real time IoT data with 3D representation of buildings. The architecture is validated in a case study creating a digital twin platform that enables users to define the thermal dynamics of buildings based on physical laws and real data, thus facilitating informed decision making for the best heating energy optimization strategy. Also, the created user interface enables stakeholders such as facility managers, energy providers or governing bodies to analyse, compare and evaluate buildings thermal dynamics without extensive expertise or time resources.
This paper addresses the prevalent issue of label shift in an online setting with missing labels, where data distributions change over time and obtaining timely labels is challenging. While existing methods primarily focus on adjusting or updating the final layer of a pre-trained classifier, we explore the untapped potential of enhancing feature representations using unlabeled data at test-time. Our novel method, Online Label Shift adaptation with Online Feature Updates (OLS-OFU), leverages self-supervised learning to refine the feature extraction process, thereby improving the prediction model. Theoretical analyses confirm that OLS-OFU reduces algorithmic regret by capitalizing on self-supervised learning for feature refinement. Empirical studies on various datasets, under both online label shift and generalized label shift conditions, underscore the effectiveness and robustness of OLS-OFU, especially in cases of domain shifts.
This paper explores Physical-Layer Security (PLS) in Flexible Duplex (FlexD) networks, considering scenarios involving eavesdroppers. Our investigation revolves around the intricacies of the sum secrecy rate maximization problem, particularly when faced with coordinated and distributed eavesdroppers employing a Minimum Mean Square Error (MMSE) receiver. Our contributions include an iterative classical optimization solution and an unsupervised learning strategy based on Graph Neural Networks (GNNs). To the best of our knowledge, this work marks the initial exploration of GNNs for PLS applications. Additionally, we extend the GNN approach to address the absence of eavesdroppers' channel knowledge. Extensive numerical simulations highlight FlexD's superiority over Half-Duplex (HD) communications and the GNN approach's superiority over the classical method in both performance and time complexity.
Background: Eating disorders are increasingly prevalent, and social networks offer valuable information. Objective: Our goal was to identify efficient machine learning models for categorizing tweets related to eating disorders. Methods: Over three months, we collected tweets about eating disorders. A 2,000-tweet subset was labeled for: (1) being written by individuals with eating disorders, (2) promoting eating disorders, (3) informativeness, and (4) scientific content. Both traditional machine learning and deep learning models were employed for classification, assessing accuracy, F1 score, and computational time. Results: From 1,058,957 collected tweets, transformer-based bidirectional encoder representations achieved the highest F1 scores (71.1%-86.4%) across all four categories. Conclusions: Transformer-based models outperform traditional techniques in classifying eating disorder-related tweets, though they require more computational resources.
Data annotation is an essential step for constructing new datasets. However, the conventional approach of data annotation through crowdsourcing is both time-consuming and expensive. In addition, the complexity of this process increases when dealing with low-resource languages owing to the difference in the language pool of crowdworkers. To address these issues, this study proposes an autonomous annotation method by utilizing large language models, which have been recently demonstrated to exhibit remarkable performance. Through our experiments, we demonstrate that the proposed method is not just cost-efficient but also applicable for low-resource language annotation. Additionally, we constructed an image captioning dataset using our approach and are committed to open this dataset for future study. We have opened our source code for further study and reproducibility.
We present an active learning algorithm for the problem of body schema learning, i.e. estimating a kinematic model of a serial robot. The learning process is done online using Recursive Least Squares (RLS) estimation, which outperforms gradient methods usually applied in the literature. In addiction, the method provides the required information to apply an active learning algorithm to find the optimal set of robot configurations and observations to improve the learning process. By selecting the most informative observations, the proposed method minimizes the required amount of data. We have developed an efficient version of the active learning algorithm to select the points in real-time. The algorithms have been tested and compared using both simulated environments and a real humanoid robot.
Multiscale problems are challenging for neural network-based discretizations of differential equations, such as physics-informed neural networks (PINNs). This can be (partly) attributed to the so-called spectral bias of neural networks. To improve the performance of PINNs for time-dependent problems, a combination of multifidelity stacking PINNs and domain decomposition-based finite basis PINNs are employed. In particular, to learn the high-fidelity part of the multifidelity model, a domain decomposition in time is employed. The performance is investigated for a pendulum and a two-frequency problem as well as the Allen-Cahn equation. It can be observed that the domain decomposition approach clearly improves the PINN and stacking PINN approaches.
In Software Product Lines (SPLs) it may be difficult or even impossible to test all the products of the family because of the large number of valid feature combinations that may exist. Thus, we want to find a minimal subset of the product family that allows us to test all these possible combinations (pairwise). Furthermore, when testing a single product is a great effort, it is desirable to first test products composed of a set of priority features. This problem is called Prioritized Pairwise Test Data Generation Problem. State-of-the-art algorithms based on Integer Linear Programming for this problema are faster enough for small and medium instances. However, there exists some real instances that are too large to be computed with these algorithms in a reasonable time because of the exponential growth of the number of candidate solutions. Also, these heuristics not always lead us to the best solutions. In this work we propose a new approach based on a hybrid metaheuristic algorithm called Construct, Merge, Solve & Adapt. We compare this matheuristic with four algorithms: a Hybrid algorithm based on Integer Linear Programming ((HILP), a Hybrid algorithm based on Integer Nonlinear Programming (HINLP), the Parallel Prioritized Genetic Solver (PPGS), and a greedy algorithm called prioritized-ICPL. The analysis reveals that CMSA results in statistically significantly better quality solutions in most instances and for most levels of weighted coverage, although it requires more execution time.
Over the recent years, the emergence of large language models (LLMs) has given rise to a proliferation of domain-specific models that are intended to reflect the particularities of linguistic context and content as a correlate of the originating domain. This paper details the conception, design, training and evaluation of DAEDRA, a LLM designed to detect regulatory-relevant outcomes (mortality, ER attendance and hospitalisation) in adverse event reports elicited through passive reporting (PR). While PR is a highly cost-efficient way of eliciting information from a wide and diverse audience -- typically including not only physicians and healthcare providers but also patients, family members and other lay stakeholders --, this diversity makes PR corpora difficult to analyse. Generic language models may not capture the complex clinical dimensions while specific clinical or biomedical models may not perform well on lay reports. To evaluate the utility of a subdomain-specific language model, an adaptive training approach was adapted, wherein base language model candidates were evaluated on a subset of the corpus, and the best performer was trained on the entire corpus. This yielded a small but significant improvement in $F_1$ (+1%), precision (+2.5%) and recall (+3.8%), at a relatively low training cost and a single-day training time. Subdomain-specific LLMs continue to be viable options for better results when analysing highly specialised corpora.
Denoising Diffusion Probabilistic Models (DDPMs) are a very popular class of deep generative model that have been successfully applied to a diverse range of problems including image and video generation, protein and material synthesis, weather forecasting, and neural surrogates of partial differential equations. Despite their ubiquity it is hard to find an introduction to DDPMs which is simple, comprehensive, clean and clear. The compact explanations necessary in research papers are not able to elucidate all of the different design steps taken to formulate the DDPM and the rationale of the steps that are presented is often omitted to save space. Moreover, the expositions are typically presented from the variational lower bound perspective which is unnecessary and arguably harmful as it obfuscates why the method is working and suggests generalisations that do not perform well in practice. On the other hand, perspectives that take the continuous time-limit are beautiful and general, but they have a high barrier-to-entry as they require background knowledge of stochastic differential equations and probability flow. In this note, we distill down the formulation of the DDPM into six simple steps each of which comes with a clear rationale. We assume that the reader is familiar with fundamental topics in machine learning including basic probabilistic modelling, Gaussian distributions, maximum likelihood estimation, and deep learning.