Building energy planning is a challenging task in the current mounting climate change scenario because the sector accounts for a reasonable percentage of global end-use energy consumption, with a one-fifth share of global carbon emissions. Energy planners rely on physical model-based prediction tools to conserve energy and make decisions towards decreasing energy consumption. For precise forecasting, such a model requires the collection of an enormous number of input variables, which is time-consuming because not all the parameters are easily available. Utilities are reluctant to share retrievable consumer information because of growing concerns regarding data leakage and competitive energy markets. Federated learning (FL) provides an effective solution by providing privacy preserving distributed training to relieve the computational burden and security concerns associated with centralized vanilla learning. Therefore, we aimed to comparatively analyze the effectiveness of several data-driven prediction algorithms for learning patterns from data-efficient buildings to predict the hourly consumption of the building sector in centralized and FL setups. The results provided comparable insights for predicting building energy consumption in a distributed setup and for generalizing to diverse clients. Moreover, such research can benefit energy designers by allowing them to use appropriate algorithms via transfer learning on data of similar features and to learn personalized models in meta-learning approaches.

MuSe-Stress 2022 aims at building sequence regression models for predicting valence and physiological arousal levels of persons who are facing stressful conditions. To that end, audio-visual recordings, transcripts, and physiological signals can be leveraged. In this paper, we describe the approach we developed for Muse-Stress 2022. Specifically, we engineered a new pose feature that captures the movement of human body keypoints. We also trained a Long Short-Term Memory (LSTM) network and a Transformer encoder on different types of feature sequences and different combinations thereof. In addition, we adopted a two-pronged strategy to tune the hyperparameters that govern the different ways the available features can be used. Finally, we made use of late fusion to combine the predictions obtained for the different unimodal features. Our experimental results show that the newly engineered pose feature obtains the second highest development CCC among the seven unimodal features available. Furthermore, our Transformer encoder obtains the highest development CCC for five out of fourteen possible combinations of features and emotion dimensions, with this number increasing from five to nine when performing late fusion. In addition, when searching for optimal hyperparameter settings, our two-pronged hyperparameter tuning strategy leads to noticeable improvements in maximum development CCC, especially when the underlying models are based on an LSTM. In summary, we can conclude that our approach is able to achieve a test CCC of 0.6196 and 0.6351 for arousal and valence, respectively, securing a Top-3 rank in Muse-Stress 2022.

The amalgamation of the Internet of Things (IoT) and federated learning (FL) is leading the next generation of data usage due to the possibility of deep learning with data privacy preservation. The FL architecture currently assumes labeled data samples from a client for supervised classification, which is unrealistic. Most research works in the literature focus on local training, update receiving, and global model updates. However, by principle, the labeling must be performed on the client side because the data samples cannot leave the source under the FL principle. In the literature, a few works have proposed methods for unlabeled data for FL using “class-prior probabilities” or “pseudo-labeling”. However, these methods make either unrealistic or uncommon assumptions, such as knowing class-prior probabilities are impractical or unavailable for each classification task and even more challenging in the IoT ecosystem. Considering these limitations, we explored the possibility of performing federated learning with unlabeled data by providing a clustering-based method of labeling the sample before training or federation. The proposed work will be suitable for every type of classification task. We performed different experiments on the client by varying the labeled data ratio, the number of clusters, and the client participation ratio. We achieved accuracy rates of $87\%$ and $90\%$ by using $0.01$ and $0.03$ of the truth labels, respectively.

Modern vehicles have communication protocols that are vulnerable to cyberattacks. A lack of authentication and encryption makes Controller Area Network (CAN) bus messages susceptible to injection attacks, spoofing, and time-based security threats. Existing intrusion detection systems (IDS) based on deep learning models cannot extract fine-grained features from CAN message frames. These models also have a limited ability to detect stealthy attacks and require significant computational resources. Additionally, large transformer-based models struggle to balance detection accuracy with the resource constraints of in-vehicle electronic control units (ECUs), hindering their practical deployment. Therefore, we propose a lightweight, deep-learning-based approach that can effectively detect anomalies with high accuracy. Our approach combines feature engineering, payload splitting, and label matching with a lightweight CNN-LSTM model that accurately flags intrusions while maintaining real-time efficiency. Feature engineering and payload splitting maximize knowledge extraction from CAN messages. Label matching analyzes flags for each set of attributes using a supervised CNN-LSTM-based detector to capture anomalies missed by time-window-based labeling approaches. The proposed approach is evaluated on two distinct datasets: HCRL car-hacking and ORNL ROAD, in both multiclass and binary attack-detection settings. We compare our proposed architecture against multiple baseline approaches, including rule-based checks and deep learning-based models. Experimental results demonstrate that the CNN–LSTM model achieves near-perfect detection performance, attaining up to 100% accuracy, F1-score, and ROC-AUC on the Car-Hacking dataset, while consistently outperforming CNN, LSTM, ANN, hybrid, and rule-based baselines. On the ROAD dataset, the proposed model achieves F1-scores of 0.9976 for masquerade attacks and 0.9578 for fabrication attacks, with robust performance across combined attack scenarios. Notably, the CNN–LSTM architecture requires significantly fewer parameters (as low as 8.4%) compared to transformer-based models, while delivering comparable or superior detection accuracy. These results highlight the effectiveness and practicality of the proposed IDS for real-time, resource-constrained in-vehicle deployment.

In this paper, we discuss the way we addressed the 2024 MuSe-Perception challenge, which aims to automatically recognize and quantify social attributes of CEOs using multimodal data. We investigated five different approaches: (1) optimizing a basic RNN and Transformer encoder model, (2) adopting the xLSTM architecture for improved long-term dependency modeling, (3) extracting text features using pre-trained language models, (4) grouping similar social attributes for joint learning, and (5) incorporating novel fusion methods. The experimental results on the development set showed that our different approaches excelled for different attributes, with multimodal methods generally outperforming unimodal methods, and text feature extraction notably improving the prediction of a subset of the agentive attributes. For the test set, our most effective strategy achieved a mean Pearson correlation coefficient of 0.35 by combining the highest values from four different approaches. This outcome points to the existence of a substantial gap between development and test set effectiveness, indicating limitations in model generalization and calling for further research to improve the accuracy and reliability of methods for social attribute prediction.