Predictive maintenance involves collecting data from machines and using algorithms to analyze the machine’s condition or determine if the machine requires maintenance or repairs. This work presents a clustering-based algorithm for predictive maintenance that detects potential faults and gradual deterioration for IoT-based buses. It demonstrates that predictive maintenance enhances cost and time efficiency and improves user safety by enabling preemptive maintenance actions. While the predictive models implemented in this article focus on the cooling and engine torque systems, the methodology proposed is flexible and can be extended to other subsystems. To mitigate the problem of insufficient data, this work also generates synthetic datasets to simulate normal buses and buses with potential faults. Experiments on synthetic datasets simulating 78 buses deliver high-quality clusters with silhouette scores as high as 0.99 (cooling system) and 0.88 (engine system). Furthermore, the clusters identify the faulty components with an accuracy of 100%, that is, all the buses with potential faults were detected successfully. Predictive maintenance frameworks usually require large volumes of labeled data and suffer from imbalance issues; however, the proposed methodology in this article delivers highly accurate results even in the absence of large volumes of labeled data while being robust against imbalanced cases. Overall, this work contributes to predictive maintenance by presenting an efficient and practical solution that ensures the reliability and safety of transportation systems.

Federated Learning (FL) enables collaborative training of Machine Learning (ML) models while maintaining user data privacy. However, leaked model updates can reveal private training data. Existing solutions using additive secret sharing introduce intermediary servers, increasing complexity and communication overhead, and often lack privacy guarantees. We propose AddShare+, which enhances efficiency and scalability by creating additive shares for a subset of model weight parameters and using the Elliptic Curve Integrated Encryption Scheme (ECIES) for faster, lighter model encryption. By sampling and splitting a percentage of local weight parameters, AddShare+ reduces computation and communication costs while maintaining model accuracy. We implemented and evaluated AddShare+ on multiple datasets, comparing it with baseline approaches including FedAvg, SCOTCH, FedShare, and AddShare. Results demonstrate that AddShare+ maintains accuracy while significantly reducing running time per round. Notably, sharing as low as 25% of model weights decreases bandwidth demands by over 5x while preserving accuracy within 0.05 % of the full model. Our empirical results demonstrate significant reductions in running time per round with strong privacy guarantees, highlighting the potential of lightweight partial sharing solutions for privacy-preserving FL in resource-constrained environments, paving the way for more efficient and secure collaborative learning systems.

This paper introduces the Subspace Rotation Algorithm (SRA) for training the Restricted Hopfield Network (RHN) as an auto-associative memory. SRA is a gradient-free subspace tracking method based on Singular Value Decomposition (SVD) to update the weight matrix. Despite having slightly worse time complexity than Back-propagation (BP) theoretically, in practice, SRA completes training faster since it requires fewer iterations to converge. Comparative analysis with BP for training RHN reveals that SRA consistently reaches the optimal solution, whereas BP fails to achieve comparable performance if the weight initialization is not within the appropriate basin of attraction. Experiments involving the memorization of 10, 50, and 100 patterns from the MNIST dataset show that RHN trained with SRA exhibits better robustness to noisy and corrupted patterns compared to RHN trained with BP. These findings suggest that SRA offers a more reliable and effective method for training RHNs in applications needing high tolerance to input distortions.

Environmental sustainability is crucial for ensuring the long-term health and well-being of our planet and its in-habitants. Precise navigation of autonomous and semi-autonomous vehicles in agricultural usage, therefore, becomes a crucial com-ponent in ensuring such sustainability. It is no surprise that the use of visible light communication (VLC) in navigation is steadily expanding, driven primarily by its remarkable precision in indoor environments. VLC has significant potential as a su-perior alternative to traditional navigation systems, including the global positioning system (GPS), particularly in dynamic outdoor environments. However, despite its indoor success, addressing the challenges posed by ambient light remains critical, particularly in outdoor scenarios where natural lighting conditions vary substan-tially. This article investigates the complex relationship between ambient light and outdoor VLC technology. The study aims to quantify the impact of ambient light on the receiving power level at varying distances in outdoor VLC systems by employing advanced simulation and modelling techniques. It provides a detailed analysis of the effects of sunlight, offering valuable insights for optimizing the positioning of light transmitters and receivers to enhance performance and reliability in real-world applications. The results discussed have significant implications for advancing outdoor VLC technology. They provide a pathway for creating more resilient and efficient navigation systems that function seamlessly across various environmental conditions.

Agriculture and Agri-Food Canada, in its unwavering commitment to sustainable agriculture, has launched a program to reduce nitrous oxide (N 2 O) emissions from fertilizer utilization in farming practices. This initiative is a response to the pressing environmental and climate challenges we face. To achieve our goal, we must delve into the mechanism of N 2 O emission by measuring and predicting the flux of N 2 O. This study proposes a novel architecture for neural network models, namely the agriculture-informed neural network (AINN) model, consisting of recurrent neural networks and a process-based ecosystem model, the Dynamic Land Ecosystem Model (DLEM), to predict N 2 O emissions from farming. During the 2021 and 2022 growing seasons, field data on the flux of N 2 O, soil temperature, and soil moisture were collected. However, the amount of nitrate in the soil was missing since collecting accurate data on nitrate quantities from the soil was challenging. Therefore, assumptions about the nitrate quantity in the soil were made when training and testing AINN with the data collected from the 2021 and 2022 growing seasons. In 2024, from January to April, an indoor experiment under controlled conditions was successfully executed to collect data on nitrate quantity in the soil. This experiment demonstrated that nitrate quantity is an essential factor for predicting the emission of N 2 O. To demonstrate the versatility of the AINN across various neural networks, we conduct a comprehensive comparison with four state-of-the-art models: multilayer perceptron, convolutional neural network, long short-term memory, and Transformer. Our experiment and simulation results unequivocally demonstrate that the performance of AINN is superior to single neural network models. The DLEM component of the AINN acts as a regularizer, facilitating the training process of the AINN. This mathematical formulation transforms the problem of N 2 O emission into a constrained optimization issue, minimizing the explicit objective function and satisfying the constraints of the parameters fed into the DLEM in the AINN. The empirical results show that by incorporating information from the agricultural field, the AINN significantly reduces the generalization error compared to the corresponding neural network, underscoring its potential to revolutionize the field of neural network modeling.

Input cost reductions and yield optimization can be done using yield precision maps created by machine learning models to address the increase in food demand predicted by 2050. However, without taking into account the spatial structure of the data, the precision map’s accuracy evaluation assessment runs the risk of being overly optimistic. In the current work, a corn yield prediction study was conducted, and the predictive abilities of two vegetation indices (VIs) and five spectral bands for a single image acquisition date were evaluated. We also examined the impacts of image spatial and spectral resolution on model performance. We used a Canadian smart farm’s yield data, multispectral (MS) and red-green-blue (RGB) imagery captured by unmanned aerial vehicles (UAVs), and we trained deep neural networks (DNN), random forest (RF), and linear regression (LR) models using standard cross-validation and spatial cross-validation approaches. We found that multi-band datasets led to better performance than single-VI datasets. MS imagery led to generally better performance than RGB imagery. High spatial resolution imagery led to better performance than lower spatial resolution imagery. RF was the best performing model while LR was the worst. The choice of RF’s hyperparameters had more of an impact on performance when the number of features was small and less of an impact when the number of features was large or when the input dataset had a lot of spatial structure.

As the global population grows rapidly, ensuring food security has become a challenge. Climate change on the other hand has led to increased frequency and intensity of weather conditions, posing significant risks to agriculture production. An accurate yield prediction system plays a crucial role in addressing the challenge by enabling effective resource allocation, optimizing agriculture practices, and reducing risk. By accurately estimating end-of-season yield in advance, farmers can take timely proactive measures for risk mitigation and yield improvement. Current approaches suffer from imprecision, an incapacity to capture intricate nonlinear connections and the challenge of accounting for spatial-temporal variations. The current work proposes an end-to-end framework using 3 dimensional CNN models and compares different strategies to improve prediction performance. The data was collected from farms located in Ottawa, Ontario, where predominantly corn (measured in bushels per acre, bu/ac) is cultivated from the 2021 growing season divided into Early, Mid, and Later stages. Two CNN models (a) 2D CNN and (b) 3D CNN were tested, where the widely used 2D CNN model was used as a baseline. The findings demonstrated that the 3D CNN model, which also incorporates temporal features(crop changes over time), outperformed the 2D CNN model, which exclusively focuses on spatial characteristics. Overall, the 3D CNN model with stacked Early and growing season images was able to achieve a Mean Absolute Percentage Error of 15.18% and a Root Mean Square Error of 17.63 bu/ac.

In recent years, big data produced by the Internet of Things has enabled new kinds of useful applications. One such application is monitoring a fleet of vehicles in real time to predict their remaining useful life. The consensus self-organized models (COSMO) approach is an example of a predictive maintenance system. The present work proposes a novel Internet of Things based architecture for predictive maintenance that consists of three primary nodes: the vehicle node, the server leader node, and the root node, which enable on-board vehicle data processing, heavy-duty data processing, and fleet administration, respectively. A minimally viable prototype of the proposed architecture was implemented and deployed to a local bus garage in Gatineau, Canada. The present work proposes improved consensus self-organized models (ICOSMO), a fleet-wide unsupervised dynamic sensor selection algorithm. To analyze the performance of ICOSMO, a fleet simulation was implemented. The J1939 data gathered from a hybrid bus was used to generate synthetic data in the simulations. Simulation results that compared the performance of the COSMO and ICOSMO approaches revealed that in general ICOSMO improves the average area under the curve of COSMO by approximately 1.5% when using the Cosine distance and 0.6% when using Hellinger distance.

Identifying anomalous samples from highly complex and unstructured data is a crucial but challenging task in a variety of intelligent systems. In this paper, we present a novel deep anomaly detection framework named AnoDM (standing for Anomaly detection based on unsupervised Disentangled representation learning and Manifold learning). The disentanglement learning is currently implemented by β-VAE for automatically discovering interpretable factorized latent representations in a completely unsupervised manner. The manifold learning is realized by t-SNE for projecting the latent representations to a 2D map. We define a new anomaly score function by combining β-VAE's reconstruction error in the raw feature space and local density estimation in the t-SNE space. AnoDM was evaluated on both image and time-series data and achieved better results than models that use just one of the two measures and other deep learning methods.

Recent research has investigated approaches and models to produce optimal results in social recommendation systems (SRSs) particularly in text-based form. The aim is to analyze the user generated-content (UGC) to suggest appropriate recommendations to interested users. However, users are often not satisfied with the initial recommendations because some models do not elicit their preferences at the beginning of the interaction nor do they understand their actual needs. In this paper, we propose a real-time SRSs called ChatWithRec that aims to improve the accuracy of recommendations by analyzing the user's contextual conversation dynamically, detect the topic, and then match it with a suitable advertisement. We used the Latent Dirichlet Allocation topic model (LDA) to analyze the user's conversation and perceive topics. We evaluated our system by applying several metrics like coherence, and F-score to evaluate the performance of ChatWithRec recommendation system. The results are encouraging, indicating that the system is fast, satisfies users by getting exactly what they seek in their conversation flow.

In this paper, we use a recently developed dynamic model for urban traffic flow subject to practical constraint characteristics of intersections equipped with traffic light. We define an objective functional based on the analytical expressions for traffic throughput, congestion, and drivers’ waiting time at an intersection. Following this, an optimization problem is formulated and an algorithm is presented based on the principle of optimality due to Bellman. The solution, if implemented, is expected to improve throughput, reduce congestion, avoid traffic jams, and promote driver satisfaction. The system is simulated with a series of numerical experiments and the corresponding optimization problems are solved using the proposed algorithm. The optimal feedback control laws independent of initial state are acquired, and the optimal cost is found according to any given initial condition. It is believed that this dynamic model would be potentially applicable for the real-time adaptive traffic control system.

This paper presents a new approach for speech feature enhancement in the log-spectral domain for noisy speech recognition. A switching linear dynamic model (SLDM) is explored as a parametric model for the clean speech distribution. Each multivariate linear dynamic model (LDM) is associated with the hidden state of a hidden Markov model (HMM) as an attempt to describe the temporal correlations among adjacent frames of speech features. The state transition on the Markov chain is the process of activating a different LDM or activating some of them simultaneously by different probabilities generated by the HMM. Rather than holding a transition probability for the whole process, a connectionist model is employed to learn the time variant transition probabilities. With the resulting SLDM as the speech model and with a model for the noise, speech and noise are jointly tracked by means of switching Kalman filtering. Comprehensive experiments are carried out using the Aurora2 database to evaluate the new algorithm. The results show that the new SLDM approach can further improve the speech feature enhancement performance in terms of noise-robust recognition accuracy, since the transition probabilities among the LDMs can be described more precisely at each time point.

This paper presents an approach to enhance speech feature estimation in the log spectral domain under noisy environments. A higher-order switching linear dynamic model (SLDM) is explored as a parametric model for the clean speech distribution, which enforces a state transition in the feature space and captures the smooth time evolution of speech conditioned on the state sequence. The clean speech components are estimated by means of an Interacting Multiple Model (IMM) algorithm. Our experimental results show that increasing the order of the linear dynamic model in the SLDM and the introduction of transition probabilities among the linear dynamic models can improve the performance of SLDM systems in feature compensation for robust speech recognition.

This paper examines the use of different feature sets and normalization techniques for a signature and password verifier. The verifier made use of the dynamic time warping (DTW) algorithm. Features that incorporated the pen velocity were found to be the strongest performers in tests against informed forgeries. Overall, password verification did not perform as well as signature verification. On average, the equal error rate for passwords was 2.7% higher than for signatures. Most signature and password features achieved their best performance when they were power-normalized, although normalization in both time and power was sometimes beneficial as well

This letter presents a new approach to enhance speech feature estimation in the log spectral domain under noisy environments. A mixture of linear dynamic models with an architecture similar to the so-called mixture of experts (ME) is investigated to describe the clean speech feature distribution parametrically. Switching Kalman filters are adapted to the proposed model, and they estimate the clean speech components by means of a generalized pseudo-Bayesian (GPB) algorithm. Experimental results suggest that compared with previous methods, the proposed approach can be more powerful to compensate the noisy speech features for robust speech recognition

In this paper, a switching Kalman filter (SKF) with a generalized pseudo Bayesian (GPB) algorithm of order 1 is applied to the problem of speech enhancement. It is proposed to use the masking properties of human auditory systems as a perceptual post-filter concatenated with the GPB algorithm. Experiments show that the proposed algorithm can achieve an improvement both in terms of speech quality (PESQ score, ITU-T P.862) and of word recognition rate at low SNR.

In some communication systems, the transmitted signal is contaminated by impulsive noise with a non-Gaussian distribution. Non-Gaussian noise causes significant performance degradation to communication receivers. In this paper, we apply a recurrent neural equalizer to impulsive noise channels, for which the performance of neural network equalizers has never been evaluated. This new application is motivated from the fact that the unscented Kalman filter (UKF), which is suited for training of the recurrent neural equalizer, provides a higher accuracy than the extended Kalman filter (EKF) in capturing the statistical characteristics for non-Gaussian random variables. The performance of the recurrent neural equalizer is evaluated for impulsive noise channels through Monte Carlo simulations. The results support the superiority of the UKF to the EKF in compensating the effect of non-Gaussian impulsive noise.

This letter uses the twisted-pair common-mode signal as the input of an adaptive canceller to remove some wideband crosstalk in a digital subscriber line (xDSL) differential signal. Simulations using simple crosstalk and line balance models show the potential benefits of the canceller to improve the signal-to-noise ratio of an xDSL channel.

Real-time recurrent learning (RTRL), commonly employed for training a fully connected recurrent neural network (RNN), has a drawback of slow convergence rate. In the light of this deficiency, a decision feedback recurrent neural equalizer (DFRNE) using the RTRL requires long training sequences to achieve good performance. In this paper, extended Kalman filter (EKF) algorithms based on the RTRL for the DFRNE are presented in state-space formulation of the system, in particular for complex-valued signal processing. The main features of global EKF and decoupled EKF algorithms are fast convergence and good tracking performance. Through nonlinear channel equalization, performance of the DFRNE with the EKF algorithms is evaluated and compared with that of the DFRNE with the RTRL.

Efficient design of suitable public key cryptographic algorithms is one of the most important problems facing their use in communication systems. An emerging public key cryptosystem that promises to be extremely useful for devices built on embedded systems which are resource constrained in memory, space and processing power is that of hyperelliptic curve cryptosystem (HECC). This work outlines an FPGA implementation of a HEC coprocessor which is based on projective and mixed coordinate representations of the curves. A transformation of variables of the curves is also suggested to improve the operating time. Numerical results are provided that shows the improvements offered by these implementations in terms of space and operation time.

Crosstalk originating from multiple high-speed data services in the same telephone bundle is a limiting factor for the maximum bit rate, the loop length and the number of data services that a bundle can support. Due to the nature of twisted-pairs, external interferences (including crosstalk) mostly couple to the twisted-pair line in common mode, and then leaks to differential mode due to line imperfect balance. The result is a degradation of the received differential signal quality. This paper uses the common mode signal as a reference to an adaptive wideband crosstalk canceller, as an attempt to remove the effect of crosstalk on the differential signal. Simulation results show the potential benefits of using this technique to reduce the crosstalk levels.

The ability of multilayer perceptron (MLP) and radial basis function (RBF) neural networks to predict the future output of chaotic and non-chaotic nonlinear dynamical systems (NDS) is analyzed. Static (i.e., feedforward) MLP and RBF neural nets (NN) are trained using a NDS with a stable attractor. The capabilities and limitations of each net architecture in terms of prediction accuracy are discussed. Emphasis is also placed on identifying the training problems for each net structure and relating these to their inherent capabilities and limitations. Static and locally recurrent RBF NN are also trained on a NDS with a chaotic attractor (i.e., the Lorenz attractor). The prediction ability of a static net structure for NDS with stable attractors and for NDS with a chaotic attractor are compared. The impact of adding feedback to the RBF neurons in terms of prediction ability is also analyzed. Training problems for each net structure are also discussed.