Search

Humans learn from a lot of information sources to make decisions. Once this information is learned in the brain, spatio-temporal associations are made, connecting all these sources (variables) in space and time represented as brain connectivity. In reality, to make a decision, we usually have only part of the information, either as a limited number of variables, limited time to make the decision, or both. The brain functions as a spatio-temporal associative memory. Inspired by the ability of the human brain, a brain-inspired spatio-temporal associative memory was proposed earlier that utilized the NeuCube brain-inspired spiking neural network framework. Here we applied the STAM framework to develop STAM for neuroimaging data, on the cases of EEG and fMRI, resulting in STAM-EEG and STAM-fMRI. This paper showed that once a NeuCube STAM classification model was trained on a complete spatio-temporal EEG or fMRI data, it could be recalled using only part of the time series, or/and only part of the used variables. We evaluated both temporal and spatial association and generalization accuracy accordingly. This was a pilot study that opens the field for the development of classification systems on other neuroimaging data, such as longitudinal MRI data, trained on complete data but recalled on partial data. Future research includes STAM that will work on data, collected across different settings, in different labs and clinics, that may vary in terms of the variables and time of data collection, along with other parameters. The proposed STAM will be further investigated for early diagnosis and prognosis of brain conditions and for diagnostic/prognostic marker discovery. [ABSTRACT FROM AUTHOR]

Vision-based perception modules are increasingly deployed in many applications, especially autonomous vehicles and intelligent robots. These modules are being used to acquire information about the surroundings and identify obstacles. Hence, accurate detection and classification are essential to reach appropriate decisions and take appropriate and safe actions at all times. Current studies have demonstrated that "printed adversarial attacks", known as physical adversarial attacks, can successfully mislead perception models such as object detectors and image classifiers. However, most of these physical attacks are based on noticeable and eye-catching patterns for generated perturbations making them identifiable/detectable by the human eye, in-field tests, or in test drives. In this paper, we propose a camera-based inconspicuous adversarial attack (AdvRain) capable of fooling camera-based perception systems over all objects of the same class. Unlike mask-based FakeWeather attacks that require access to the underlying computing hardware or image memory, our attack is based on emulating the effects of a natural weather condition (i.e., Raindrops) that can be printed on a translucent sticker, which is externally placed over the lens of a camera whenever an adversary plans to trigger an attack. Note, such perturbations are still inconspicuous in real-world deployments and their presence goes unnoticed due to their association with a natural phenomenon. To accomplish this, we develop an iterative process based on performing a random search aiming to identify critical positions to make sure that the performed transformation is adversarial for a target classifier. Our transformation is based on blurring predefined parts of the captured image corresponding to the areas covered by the raindrop. We achieve a drop in average model accuracy of more than 45 % and 40 % on VGG19 for ImageNet dataset and Resnet34 for Caltech-101 dataset, respectively, using only 20 raindrops. [ABSTRACT FROM AUTHOR]

Digital holographic microscopy (DHM) is a valuable technique for investigating the optical properties of samples through the measurement of intensity and phase of diffracted beams. However, DHMs are constrained by Lagrange invariance, compromising the spatial bandwidth product (SBP) which relates resolution and field of view. Synthetic aperture DHM (SA-DHM) was introduced to overcome this limitation, but it faces significant challenges such as aberrations in synthesizing the optical information corresponding to the steering angle of incident wave. This paper proposes a novel approach utilizing deep neural networks (DNNs) for compensating aberrations in SA-DHM, extending the compensation scope beyond the numerical aperture (NA) of the objective lens. The method involves training a DNN from diffraction patterns and Zernike coefficients through a circular aperture, enabling effective aberration compensation in the illumination beam. This method makes it possible to estimate aberration coefficients from the only part of the diffracted beam cutoff by the circular aperture mask. With the proposed technique, the simulation results present improved resolution and quality of sample images. The integration of deep neural networks with SA-DHM holds promise for advancing microscopy capabilities and overcoming existing limitations. [ABSTRACT FROM AUTHOR]

Recent successes in deep learning have inspired researchers to apply deep neural networks to Acoustic Event Classification (AEC). While deep learning methods can train effective AEC models, they are susceptible to overfitting due to the models' high complexity. In this paper, we introduce EnViTSA, an innovative approach that tackles key challenges in AEC. EnViTSA combines an ensemble of Vision Transformers with SpecAugment, a novel data augmentation technique, to significantly enhance AEC performance. Raw acoustic signals are transformed into Log Mel-spectrograms using Short-Time Fourier Transform, resulting in a fixed-size spectrogram representation. To address data scarcity and overfitting issues, we employ SpecAugment to generate additional training samples through time masking and frequency masking. The core of EnViTSA resides in its ensemble of pre-trained Vision Transformers, harnessing the unique strengths of the Vision Transformer architecture. This ensemble approach not only reduces inductive biases but also effectively mitigates overfitting. In this study, we evaluate the EnViTSA method on three benchmark datasets: ESC-10, ESC-50, and UrbanSound8K. The experimental results underscore the efficacy of our approach, achieving impressive accuracy scores of 93.50%, 85.85%, and 83.20% on ESC-10, ESC-50, and UrbanSound8K, respectively. EnViTSA represents a substantial advancement in AEC, demonstrating the potential of Vision Transformers and SpecAugment in the acoustic domain. [ABSTRACT FROM AUTHOR]

In this paper, we continue the research cycle on the properties of convolutional neural network-based image recognition systems and ways to improve noise immunity and robustness. Currently, a popular research area related to artificial neural networks is adversarial attacks. The adversarial attacks on the image are not highly perceptible to the human eye, and they also drastically reduce the neural network's accuracy. Image perception by a machine is highly dependent on the propagation of high frequency distortions throughout the network. At the same time, a human efficiently ignores high-frequency distortions, perceiving the shape of objects as a whole. We propose a technique to reduce the influence of high-frequency noise on the CNNs. We show that low-pass image filtering can improve the image recognition accuracy in the presence of high-frequency distortions in particular, caused by adversarial attacks. This technique is resource efficient and easy to implement. The proposed technique makes it possible to measure up the logic of an artificial neural network to that of a human, for whom high-frequency distortions are not decisive in object recognition. [ABSTRACT FROM AUTHOR]

Dynamic neural networks (DNNs) are a type of artificial neural network (ANN) designed to work with sequential data where context in time is important. Unlike traditional static neural networks that process data in a fixed order, dynamic neural networks use information about past inputs, which is important if the dynamic of a certain process is emphasized. They are commonly used in natural language processing, speech recognition, and time series prediction. In industrial processes, their use is interesting for the prediction of difficult-to-measure process variables. In an industrial isomerization process, it is crucial to measure the quality attributes that affect the octane number of gasoline. Process analyzers commonly used for this purpose are expensive and subject to failure. Therefore, to achieve continuous production in the event of a malfunction, mathematical models for estimating product quality attributes are imposed as a solution. In this paper, mathematical models were developed using dynamic recurrent neural networks (RNNs), i.e., their subtype of a long short-term memory (LSTM) architecture. The results of the developed models were compared with the results of several types of other data-driven models developed for an isomerization process, such as multilayer perceptron (MLP) artificial neural networks, support vector machines (SVM), and dynamic polynomial models. The obtained results are satisfactory, suggesting a good possibility of application. [ABSTRACT FROM AUTHOR]

Artificial intelligence (AI) radar technology offers several advantages over other technologies, including low cost, privacy assurance, high accuracy, and environmental resilience. One challenge faced by AI radar technology is the high cost of equipment and the lack of radar datasets for deep-learning model training. Moreover, conventional radar signal processing methods have the obstacles of poor resolution or complex computation. Therefore, this paper discusses an innovative approach in the integration of radar technology and machine learning for effective surveillance systems that can surpass the aforementioned limitations. This approach is detailed into three steps: signal acquisition, signal processing, and feature-based classification. A hardware prototype of the signal acquisition circuitry was designed for a Continuous Wave (CW) K-24 GHz frequency band radar sensor. The collected radar motion data was categorized into non-human motion, human walking, and human walking without arm swing. Three signal processing techniques, namely short-time Fourier transform (STFT), mel spectrogram, and mel frequency cepstral coefficients (MFCCs), were employed. The latter two are typically used for audio processing, but in this study, they were proposed to obtain micro-Doppler spectrograms for all motion data. The obtained micro-Doppler spectrograms were then fed to a simplified 2D convolutional neural networks (CNNs) architecture for feature extraction and classification. Additionally, artificial neural networks (ANNs) and 1D CNN models were implemented for comparative analysis on various aspects. The experimental results demonstrated that the 2D CNN model trained on the MFCC feature outperformed the other two methods. The accuracy rate of the object classification models trained on micro-Doppler features was 97.93%, indicating the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

Very recently, a different generalization of real-valued neural networks (RVNNs) to multidimensional domains beside the complex-valued neural networks (CVNNs), quaternion-valued neural networks (QVNNs), and Clifford-valued neural networks (ClVNNs) has appeared, namely octonion-valued neural networks (OVNNs), which are not a subset of ClVNNs. They are defined on the octonion algebra, which is an 8D algebra over the reals, and is also the only other normed division algebra that can be defined over the reals beside the complex and quaternion algebras. On the other hand, fractional-order neural networks (FONNs) have also been very intensively researched in the recent past. Thus, the present work combines FONNs and OVNNs and puts forward a fractional-order octonion-valued neural network (FOOVNN) with neutral-type, time-varying, and distributed delays, a very general model not yet discussed in the literature, to our awareness. Sufficient criteria expressed as linear matrix inequalities (LMIs) and algebraic inequalities are deduced, which ensure the asymptotic and Mittag–Leffler synchronization properties of the proposed model by decomposing the OVNN system of equations into a real-valued one, in order to avoid the non-associativity problem of the octonion algebra. To accomplish synchronization, we use two different state feedback controllers, two different types of Lyapunov-like functionals in conjunction with two Halanay-type lemmas for FONNs, the free-weighting matrix method, a classical lemma, and Young's inequality. The four theorems presented in the paper are each illustrated by a numerical example. [ABSTRACT FROM AUTHOR]

A high cognitive load can overload a person, potentially resulting in catastrophic accidents. It is therefore important to ensure the level of cognitive load associated with safety-critical tasks (such as driving a vehicle) remains manageable for drivers, enabling them to respond appropriately to changes in the driving environment. Although electroencephalography (EEG) has attracted significant interest in cognitive load research, few studies have used EEG to investigate cognitive load in the context of driving. This paper presents a feasibility study on the simulation of various levels of cognitive load through designing and implementing four driving tasks. We employ machine learning-based classification techniques using EEG recordings to differentiate driving conditions. An EEG dataset containing these four driving tasks from a group of 20 participants was collected to investigate whether EEG can be used as an indicator of changes in cognitive load. The collected dataset was used to train four Deep Neural Networks and four Support Vector Machine classification models. The results showed that the best model achieved a classification accuracy of 90.37%, utilising statistical features from multiple frequency bands in 24 EEG channels. Furthermore, the Gamma and Beta bands achieved higher classification accuracy than the Alpha and Theta bands during the analysis. The outcomes of this study have the potential to enhance the Human–Machine Interface of vehicles, contributing to improved safety. [ABSTRACT FROM AUTHOR]

Keystroke dynamics is a soft biometric based on the assumption that humans always type in uniquely characteristic manners. Previous works mainly focused on analyzing the key press or release events. Unlike these methods, we explored a novel visual modality of keystroke dynamics for human identification using a single RGB-D sensor. In order to verify this idea, we created a dataset dubbed KD-MultiModal, which contains 243.2 K frames of RGB images and depth images, obtained by recording a video of hand typing with a single RGB-D sensor. The dataset comprises RGB-D image sequences of 20 subjects (10 males and 10 females) typing sentences, and each subject typed around 20 sentences. In the task, only the hand and keyboard region contributed to the person identification, so we also propose methods of extracting Regions of Interest (RoIs) for each type of data. Unlike the data of the key press or release, our dataset not only captures the velocity of pressing and releasing different keys and the typing style of specific keys or combinations of keys, but also contains rich information on the hand shape and posture. To verify the validity of our proposed data, we adopted deep neural networks to learn distinguishing features from different data representations, including RGB-KD-Net, D-KD-Net, and RGBD-KD-Net. Simultaneously, the sequence of point clouds also can be obtained from depth images given the intrinsic parameters of the RGB-D sensor, so we also studied the performance of human identification based on the point clouds. Extensive experimental results showed that our idea works and the performance of the proposed method based on RGB-D images is the best, which achieved 99.44 % accuracy based on the unseen real-world data. To inspire more researchers and facilitate relevant studies, the proposed dataset will be publicly accessible together with the publication of this paper. [ABSTRACT FROM AUTHOR]

The lungs are critical components of the respiratory system because they allow for the exchange of oxygen and carbon dioxide within our bodies. However, a variety of conditions can affect the lungs, resulting in serious health consequences. Lung disease treatment aims to control its severity, which is usually irrevocable. The fundamental objective of this endeavor is to build a consistent and automated approach for establishing the intensity of lung illness. This paper describes MixNet-LD, a unique automated approach aimed at identifying and categorizing the severity of lung illnesses using an upgraded pre-trained MixNet model. One of the first steps in developing the MixNet-LD system was to build a pre-processing strategy that uses Grad-Cam to decrease noise, highlight irregularities, and eventually improve the classification performance of lung illnesses. Data augmentation strategies were used to rectify the dataset's unbalanced distribution of classes and prevent overfitting. Furthermore, dense blocks were used to improve classification outcomes across the four severity categories of lung disorders. In practice, the MixNet-LD model achieves cutting-edge performance while maintaining model size and manageable complexity. The proposed approach was tested using a variety of datasets gathered from credible internet sources as well as a novel private dataset known as Pak-Lungs. A pre-trained model was used on the dataset to obtain important characteristics from lung disease images. The pictures were then categorized into categories such as normal, COVID-19, pneumonia, tuberculosis, and lung cancer using a linear layer of the SVM classifier with a linear activation function. The MixNet-LD system underwent testing in four distinct tests and achieved a remarkable accuracy of 98.5% on the difficult lung disease dataset. The acquired findings and comparisons demonstrate the MixNet-LD system's improved performance and learning capabilities. These findings show that the proposed approach may effectively increase the accuracy of classification models in medicinal image investigations. This research helps to develop new strategies for effective medical image processing in clinical settings. [ABSTRACT FROM AUTHOR]

For equivalent modeling of mixed wind farms (WFs), existing clustering indicators cannot consider the complex coupling characteristics between different types of wind turbines (WTs). In this paper, a refined equivalent modeling approach based on artificial intelligence technology is proposed. Firstly, the electromechanical transient performance of mixed WFs is analyzed. The WT type, wind speed and direction, and voltage dip are considered the dominant factors affecting the external dynamic response of mixed WFs. Secondly, the equivalent node model is established, including the selection of independent and dependent variables. Then, the multiple artificial neural networks (ANNs) are trained one by one based on small sample data, to fit the nonlinear relationship between the dependent variables and the independent variables. Finally, the dynamic response of the power systems with a mixed WF is simulated in the MATLAB platform. A comparison of the errors in electromechanical phenomena demonstrates that the proposed model can reflect the external characteristics of the test mixed WF in different wind conditions and voltage dips. [ABSTRACT FROM AUTHOR]

Confronted with the climate crisis, the world is making tremendous efforts in energy transition, such as expanding renewable energy that does not emit carbon. The importance of virtual power plant (VPP) operation technology has emerged to secure grid flexibility in response to the expanding renewable energy implemented due to these efforts. Accordingly, VPPs, which include photovoltaics, wind turbines, heating, ventilation, and air conditioning (HVAC), load, and EV, have been constructed. HVAC, one of the component resources, is a system that controls and regulates temperature, humidity, and airflow. Since it responds sensitively to the building's heat capacity and changes in the external environment, it requires continuous and stable control. In this paper, we used data-based modeling to implement the HVAC required for the optimal operation of VPP. Since accurately creating an equation-based HVAC model was difficult considering building information modeling and external environment variables, we used historical HVAC operation data to perform data-based modeling. The model was implemented using nonlinear regression and machine learning, such as a support vector machine and artificial neural network. Then, the data-based HVAC and the actual HVAC operation results were comparatively analyzed based on a case study, and the model's goodness-of-fit was evaluated based on performance metrics. Model performance indicators confirmed that the ANN-based HVAC model was most similar to the actual HVAC system. [ABSTRACT FROM AUTHOR]

Ship classification using the synthetic aperture radar (SAR) images has a significant role in remote sensing applications. Aiming at the problems of excessive model parameters numbers and high energy consumption in the traditional deep learning methods for the SAR ship classification, this paper provides an energy-efficient SAR ship classification paradigm that combines spiking neural networks (SNNs) with Siamese network architecture, for the first time in the field of SAR ship classification, which is called the Siam-SpikingShipCLSNet. It combines the advantage of SNNs in energy consumption and the advantage of the idea in performances that use the Siamese neuron network to fuse the features from dual-polarized SAR images. Additionally, we migrated the feature fusion strategy from CNN-based Siamese neural networks to the SNN domain and analyzed the effects of various spiking feature fusion methods on the Siamese SNN. Finally, an end-to-end error backpropagation optimization method based on the surrogate gradient has been adopted to train this model. Experimental results tested on the OpenSARShip2.0 dataset have demonstrated the correctness and effectiveness of the proposed SAR ship classification strategy, which has the advantages of the higher accuracy, fewer parameters and lower energy consumption compared with the mainstream deep learning method of the SAR ship classification. [ABSTRACT FROM AUTHOR]

Physics-Informed Neural Networks (PINNs) improve the efficiency of data utilization by combining physical principles with neural network algorithms and thus ensure that their predictions are consistent and stable with the physical laws. PINNs open up a new approach to address inverse problems in fluid mechanics. Based on the single-relaxation-time lattice Boltzmann method (SRT-LBM) with the Bhatnagar–Gross–Krook (BGK) collision operator, the PINN-SRT-LBM model is proposed in this paper for solving the inverse problem in fluid mechanics. The PINN-SRT-LBM model consists of three components. The first component involves a deep neural network that predicts equilibrium control equations in different discrete velocity directions within the SRT-LBM. The second component employs another deep neural network to predict non-equilibrium control equations, enabling the inference of the fluid's non-equilibrium characteristics. The third component, a physics-informed function, translates the outputs of the first two networks into physical information. By minimizing the residuals of the physical partial differential equations (PDEs), the physics-informed function infers relevant macroscopic quantities of the flow. The model evolves two sub-models that are applicable to different dimensions, named the PINN-SRT-LBM-I and PINN-SRT-LBM-II models according to the construction of the physics-informed function. The innovation of this work is the introduction of SRT-LBM and discrete velocity models as physical drivers into a neural network through the interpretation function. Therefore, the PINN-SRT-LBM allows a given neural network to handle inverse problems of various dimensions and focus on problem-specific solving. Our experimental results confirm the accurate prediction by this model of flow information at different Reynolds numbers within the computational domain. Relying on the PINN-SRT-LBM models, inverse problems in fluid mechanics can be solved efficiently. [ABSTRACT FROM AUTHOR]

The process of recognising and classifying radar signals and their radiation sources is currently a key element of operational activities in the electromagnetic environment. Systems of this type, called ELINT class systems, are passive solutions that detect, process, and analyse radio-electronic signals, providing distinctive information on the identified emission source in the final stage of data processing. The data processing in the mentioned types of systems is a very sophisticated issue and is based on advanced machine learning algorithms, artificial neural networks, fractal analysis, intra-pulse analysis, unintentional out-of-band emission analysis, and hybrids of these methods. Currently, there is no optimal method that would allow for the unambiguous identification of particular copies of the same type of radar emission source. This article constitutes an attempt to analyse radar signals generated by six radars of the same type under comparable measurement conditions for all six cases. The concept of the SEI module for the ELINT system was proposed in this paper. The main aim was to perform an advanced analysis, the purpose of which was to identify particular copies of those radars. Pioneering in this research is the application of the author's algorithm for the data particle geometrical divide, which at the moment has no reference in international publication reports. The research revealed that applying the data particle geometrical divide algorithms to the SEI process concerning six copies of the same radar type allows for almost three times better accuracy than a random labelling strategy within approximately one second. [ABSTRACT FROM AUTHOR]

The rapid growth in dataset sizes in modern deep learning has significantly increased data storage costs. Furthermore, the training and time costs for deep neural networks are generally proportional to the dataset size. Therefore, reducing the dataset size while maintaining model performance is an urgent research problem that needs to be addressed. Dataset condensation is a technique that aims to distill the original dataset into a much smaller synthetic dataset while maintaining downstream training performance on any agnostic neural network. Previous work has demonstrated that matching the training trajectory between the synthetic dataset and the original dataset is more effective than matching the instantaneous gradient, as it incorporates long-range information. Despite the effectiveness of trajectory matching, it suffers from complex gradient unrolling across iterations, which leads to significant memory and computation overhead. To address this issue, this paper proposes a novel approach called Expert Subspace Projection (ESP), which leverages long-range information while avoiding gradient unrolling. Instead of strictly enforcing the synthetic dataset's training trajectory to mimic that of the real dataset, ESP only constrains it to lie within the subspace spanned by the training trajectory of the real dataset. The memory-saving advantage offered by our method facilitates unbiased training on the complete set of synthetic images and seamless integration with other dataset condensation techniques. Through extensive experiments, we have demonstrated the effectiveness of our approach. Our method outperforms the trajectory matching method on CIFAR10 by 16.7% in the setting of 1 Image/Class, surpassing the previous state-of-the-art method by 3.2%. [ABSTRACT FROM AUTHOR]

Catastrophic forgetting, which means a rapid forgetting of learned representations while learning new data/samples, is one of the main problems of deep neural networks. In this paper, we propose a novel incremental learning framework that can address the forgetting problem by learning new incoming data in an online manner. We develop a new incremental learning framework that can learn extra data or new classes with less catastrophic forgetting. We adopt the hippocampal memory process to the deep neural networks by defining the effective maximum of neural activation and its boundary to represent a feature distribution. In addition, we incorporate incremental QR factorization into the deep neural networks to learn new data with both existing labels and new labels with less forgetting. The QR factorization can provide the accurate subspace prior, and incremental QR factorization can reasonably express the collaboration between new data with both existing classes and new class with less forgetting. In our framework, a set of appropriate features (i.e., nodes) provides improved representation for each class. We apply our method to the convolutional neural network (CNN) for learning Cifar-100 and Cifar-10 datasets. The experimental results show that the proposed method efficiently alleviates the stability and plasticity dilemma in the deep neural networks by providing the performance stability of a trained network while effectively learning unseen data and additional new classes. [ABSTRACT FROM AUTHOR]

This paper presents a methodology that allows for the detection of the state of a sheet-metal-forming press, the parts being produced, their cadence, and the energy demand for each unit produced. For this purpose, only electrical measurements are used. The proposed analysis is conducted at the level of the press subsystems: main motor, transfer module, cushion, and auxiliary systems, and is intended to count, classify, and monitor the production of pressed parts. The power data are collected every 20 ms and show cyclic behavior, which is the basis for the presented methodology. A neural network (NN) based on heuristic rules is developed to estimate the press states. Then, the production period is determined from the power data using a least squares method to obtain normalized harmonic coefficients. These are the basis for a second NN dedicated to identifying the parts in production. The global error in estimating the parts being produced is under 1%. The resulting information could be handy in determining relevant information regarding the press behavior, such as energy per part, which is necessary in order to evaluate the energy performance of the press under different production conditions. [ABSTRACT FROM AUTHOR]

The automotive production sector plays a significant role in the energy consumption of all the industrial sphere, which currently represents approximately 38% of the total global energy use. Especially in production sites with several manufacturing lines working in parallel, the occurrence of failures and anomalies or sudden changes in the production volume may require a re-scheduling of the entire production process. In this regard, a digital twin of each phase of the process would give several indications about the new re-scheduled manufacture in terms of energy consumption and the control strategy to adopt. Therefore, the main goal of this paper is to propose different modeling approaches to a degreasing tank process, which is a preliminary phase at automotive production sites before the application of paint to car bodies. In detail, two different approaches have been developed: the first is a physics-based thermodynamic approach, which relies on the mass and energy balances of the system analyzed, and the second is machine learning-based, with the calibration of several artificial neural networks (ANNs). All the investigated approaches were assessed and compared, and it was determined that, for this application and with the data at our disposal, the thermodynamic approach has better prediction accuracy, with an overall mean absolute error (MAE) of 1.30 °C. Moreover, the model can be used to optimize the heat source policy of the tank, for which it has demonstrated, with historical data, an energy saving potentiality of up to 30%, and to simulate future scenarios in which, due to company constraints, a re-scheduling of the production of more work shifts is required. [ABSTRACT FROM AUTHOR]

As a safety-related application, visual systems based on deep neural networks (DNNs) in modern unmanned aerial vehicles (UAVs) show adversarial vulnerability when performing real-time inference. Recently, deep ensembles with various defensive strategies against adversarial samples have drawn much attention due to the increased diversity and reduced variance for their members. Aimed at the recognition task of remote sensing images (RSIs), this paper proposes to use a reactive-proactive ensemble defense framework to solve the security problem. In reactive defense, we fuse scoring functions of several classical detection algorithms with the hidden features and average output confidences from sub-models as a second fusion. In terms of proactive defense, we attempt two strategies, including enhancing the robustness of each sub-model and limiting the transferability among sub-models. In practical applications, the real-time RSIs are first input to the reactive defense part, which can detect and reject the adversarial RSIs. The accepted ones are then passed to robust recognition with a proactive defense. We conduct extensive experiments on three benchmark RSI datasets (i.e., UCM, AID, and FGSC-23). The experimental results show that the deep ensemble method of reactive and proactive defense performs very well in gradient-based attacks. The analysis of the applicable attack scenarios for each proactive ensemble defense is also helpful for this field. We also perform a case study with the whole framework in the black-box scenario, and the highest detection rate reaches 93.25%. Most of the adversarial RSIs can be rejected in advance or correctly recognized by the enhanced deep ensemble. This article is the first one to combine reactive and proactive defenses with a deep ensemble against adversarial attacks in the context of RSI recognition for DNN-based UAVs. [ABSTRACT FROM AUTHOR]

Pulmonary auscultation is essential for detecting abnormal lung sounds during physical assessments, but its reliability depends on the operator. Machine learning (ML) models offer an alternative by automatically classifying lung sounds. ML models require substantial data, and public databases aim to address this limitation. This systematic review compares characteristics, diagnostic accuracy, concerns, and data sources of existing models in the literature. Papers published from five major databases between 1990 and 2022 were assessed. Quality assessment was accomplished with a modified QUADAS-2 tool. The review encompassed 62 studies utilizing ML models and public-access databases for lung sound classification. Artificial neural networks (ANN) and support vector machines (SVM) were frequently employed in the ML classifiers. The accuracy ranged from 49.43% to 100% for discriminating abnormal sound types and 69.40% to 99.62% for disease class classification. Seventeen public databases were identified, with the ICBHI 2017 database being the most used (66%). The majority of studies exhibited a high risk of bias and concerns related to patient selection and reference standards. Summarizing, ML models can effectively classify abnormal lung sounds using publicly available data sources. Nevertheless, inconsistent reporting and methodologies pose limitations to advancing the field, and therefore, public databases should adhere to standardized recording and labeling procedures. [ABSTRACT FROM AUTHOR]

In recent years, deep neural networks have enabled countless innovations in the field of image classification. Encouraged by success in this field, researchers worldwide have demonstrated how to use Convolutional Neural Network techniques in medical imaging problems. In this article, the results were obtained through the use of the EfficientNet in the task of classifying 14 different diseases based on chest X-ray images coming from the NIH (National Institutes of Health) ChestX-ray14 dataset. The approach addresses dataset imbalances by introducing a custom split to ensure fair representation. Binary cross entropy loss is utilized to handle the multi-label difficulty. The model architecture comprises an EfficientNet backbone for feature extraction, succeeded by sequential layers including GlobalAveragePooling, Dense, and BatchNormalization. The main contribution of this paper is a proposed solution that outperforms previous state-of-the-art deep learning models average area under the receiver operating characteristic curve—AUC-ROC (score: 84.28%). The usage of the transfer-learning technique and traditional deep learning engineering techniques was shown to enable us to obtain such results on consumer-class GPUs (graphics processing units). [ABSTRACT FROM AUTHOR]

Simple Summary: Melanoma is the most dangerous type of skin cancer worldwide. Early detection of melanoma is crucial for better outcomes, but this often can be challenging. This research explores the use of artificial intelligence (AI) techniques combined with non-invasive imaging methods to improve melanoma detection. The authors aim to evaluate the current state of AI-based techniques using tools including dermoscopy, optical coherence tomography (OCT), and reflectance confocal microscopy (RCM). The findings demonstrate that several AI algorithms perform as well as or better than dermatologists in detecting melanoma, particularly in the analysis of dermoscopy images. This research highlights the potential of AI to enhance diagnostic accuracy, leading to improved patient outcomes. Further studies are needed to address limitations and ensure the reliability and effectiveness of AI-based techniques. Background: Melanoma, the deadliest form of skin cancer, poses a significant public health challenge worldwide. Early detection is crucial for improved patient outcomes. Non-invasive skin imaging techniques allow for improved diagnostic accuracy; however, their use is often limited due to the need for skilled practitioners trained to interpret images in a standardized fashion. Recent innovations in artificial intelligence (AI)-based techniques for skin lesion image interpretation show potential for the use of AI in the early detection of melanoma. Objective: The aim of this study was to evaluate the current state of AI-based techniques used in combination with non-invasive diagnostic imaging modalities including reflectance confocal microscopy (RCM), optical coherence tomography (OCT), and dermoscopy. We also aimed to determine whether the application of AI-based techniques can lead to improved diagnostic accuracy of melanoma. Methods: A systematic search was conducted via the Medline/PubMed, Cochrane, and Embase databases for eligible publications between 2018 and 2022. Screening methods adhered to the 2020 version of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Included studies utilized AI-based algorithms for melanoma detection and directly addressed the review objectives. Results: We retrieved 40 papers amongst the three databases. All studies directly comparing the performance of AI-based techniques with dermatologists reported the superior or equivalent performance of AI-based techniques in improving the detection of melanoma. In studies directly comparing algorithm performance on dermoscopy images to dermatologists, AI-based algorithms achieved a higher ROC (>80%) in the detection of melanoma. In these comparative studies using dermoscopic images, the mean algorithm sensitivity was 83.01% and the mean algorithm specificity was 85.58%. Studies evaluating machine learning in conjunction with OCT boasted accuracy of 95%, while studies evaluating RCM reported a mean accuracy rate of 82.72%. Conclusions: Our results demonstrate the robust potential of AI-based techniques to improve diagnostic accuracy and patient outcomes through the early identification of melanoma. Further studies are needed to assess the generalizability of these AI-based techniques across different populations and skin types, improve standardization in image processing, and further compare the performance of AI-based techniques with board-certified dermatologists to evaluate clinical applicability. [ABSTRACT FROM AUTHOR]

We present a novel information-theoretic framework, termed as TURBO, designed to systematically analyse and generalise auto-encoding methods. We start by examining the principles of information bottleneck and bottleneck-based networks in the auto-encoding setting and identifying their inherent limitations, which become more prominent for data with multiple relevant, physics-related representations. The TURBO framework is then introduced, providing a comprehensive derivation of its core concept consisting of the maximisation of mutual information between various data representations expressed in two directions reflecting the information flows. We illustrate that numerous prevalent neural network models are encompassed within this framework. The paper underscores the insufficiency of the information bottleneck concept in elucidating all such models, thereby establishing TURBO as a preferable theoretical reference. The introduction of TURBO contributes to a richer understanding of data representation and the structure of neural network models, enabling more efficient and versatile applications. [ABSTRACT FROM AUTHOR]

Stroke remains a predominant cause of mortality and disability worldwide. The endeavor to diagnose stroke through biomechanical time-series data coupled with Artificial Intelligence (AI) poses a formidable challenge, especially amidst constrained participant numbers. The challenge escalates when dealing with small datasets, a common scenario in preliminary medical research. While recent advances have ushered in few-shot learning algorithms adept at handling sparse data, this paper pioneers a distinctive methodology involving a visualization-centric approach to navigating the small-data challenge in diagnosing stroke survivors based on gait-analysis-derived biomechanical data. Employing Siamese neural networks (SNNs), our method transforms a biomechanical time series into visually intuitive images, facilitating a unique analytical lens. The kinematic data encapsulated comprise a spectrum of gait metrics, including movements of the ankle, knee, hip, and center of mass in three dimensions for both paretic and non-paretic legs. Following the visual transformation, the SNN serves as a potent feature extractor, mapping the data into a high-dimensional feature space conducive to classification. The extracted features are subsequently fed into various machine learning (ML) models like support vector machines (SVMs), Random Forest (RF), or neural networks (NN) for classification. In pursuit of heightened interpretability, a cornerstone in medical AI applications, we employ the Grad-CAM (Class Activation Map) tool to visually highlight the critical regions influencing the model's decision. Our methodology, though exploratory, showcases a promising avenue for leveraging visualized biomechanical data in stroke diagnosis, achieving a perfect classification rate in our preliminary dataset. The visual inspection of generated images elucidates a clear separation of classes (100%), underscoring the potential of this visualization-driven approach in the realm of small data. This proof-of-concept study accentuates the novelty of visual data transformation in enhancing both interpretability and performance in stroke diagnosis using limited data, laying a robust foundation for future research in larger-scale evaluations. [ABSTRACT FROM AUTHOR]

Few-shot learning is a deep learning subfield that is the focus of research nowadays. This paper addresses the research question of whether a triplet-trained Siamese network, initially designed for multi-class classification, can effectively handle multi-label classification. We conducted a case study to identify any limitations in its application. The experiments were conducted on a dataset containing Hungarian legal decisions of administrative agencies in tax matters belonging to a major legal content provider. We also tested how different Siamese embeddings compare on classifying a previously non-existing label on a binary and a multi-label setting. We found that triplet-trained Siamese networks can be applied to perform classification but with a sampling restriction during training. We also found that the overlap between labels affects the results negatively. The few-shot model, seeing only ten examples for each label, provided competitive results compared to models trained on tens of thousands of court decisions using tf-idf vectorization and logistic regression. [ABSTRACT FROM AUTHOR]

This document is a special issue of the journal Biomimetics, focusing on the design and control of bio-inspired robots. It explores various aspects of bionics in robotics, including robot design, perception, control, and decision-making, as well as incorporating neuroscience and brain science. The issue covers a wide range of topics, such as stiffness adjustment for continuum robots, biomimetic motor control, stroke rehabilitation, reinforcement learning for quadruped robots, improved spiking neural networks, energy-efficient image segmentation, kinematics analysis, synthetic nervous systems for robotic control, online running-gait generation, and bio-inspired perception and navigation for service robots. The document also discusses specific papers within the special issue that address challenges in robotic perception and navigation, legged robot control, and motion control of continuum robots and robotic arms. It concludes by announcing plans for a second special issue on related topics. [Extracted from the article]

This paper proposes facial expression recognition (FER) with the wild data set. In particular, this paper chiefly deals with two issues, occlusion and intra-similarity problems. The attention mechanism enables one to use the most relevant areas of facial images for specific expressions, and the triplet loss function solves the intra-similarity problem that sometimes fails to aggregate the same expression from different faces and vice versa. The proposed approach for the FER is robust to occlusion, and it uses a spatial transformer network (STN) with an attention mechanism to utilize specific facial region that dominantly contributes (or that is the most relevant) to particular facial expressions, e.g., anger, contempt, disgust, fear, joy, sadness, and surprise. In addition, the STN model is connected to the triplet loss function to improve the recognition rate which outperforms the existing approaches that employ cross-entropy or other approaches using only deep neural networks or classical methods. The triplet loss module alleviates limitations of the intra-similarity problem, leading to further improvement of the classification. Experimental results are provided to substantiate the proposed approach for FER, and the result outperforms the recognition rate in more practical cases, e.g., occlusion. The quantitative result provides FER results with more than 2.09 % higher accuracy compared to the existing FER results in CK+ data sets and 0.48 % higher than the accuracy of the results with the modified ResNet model in the FER2013 data set. [ABSTRACT FROM AUTHOR]

How to predict precipitation accurately and efficiently is the key and difficult problem in the field of weather forecasting. At present, we can obtain accurate meteorological data through many high-precision weather sensors and use them to forecast precipitation. However, the common numerical weather forecasting methods and radar echo extrapolation methods have insurmountable defects. Based on some common characteristics of meteorological data, this paper proposes a Pred-SF model for precipitation prediction in target areas. The model focuses on the combination of multiple meteorological modal data to carry out self-cyclic prediction and a step-by-step prediction structure. The model divides the precipitation prediction into two steps. In the first step, the spatial encoding structure and PredRNN-V2 network are used to construct the autoregressive spatio-temporal prediction network for the multi-modal data, and the preliminary predicted value of the multi-modal data is generated frame by frame. In the second step, the spatial information fusion network is used to further extract and fuse the spatial characteristics of the preliminary predicted value and, finally, output the predicted precipitation value of the target region. In this paper, ERA5 multi-meteorological mode data and GPM precipitation measurement data are used for testing to predict the continuous precipitation of a specific area for 4 h. The experimental results show that Pred-SF has strong precipitation prediction ability. Some comparative experiments were also set up for comparison to demonstrate the advantages of the combined prediction method of multi-modal data and the stepwise prediction method of Pred-SF. [ABSTRACT FROM AUTHOR]

In the fourth industrial revolution, the scale of execution for interactive applications increased substantially. These interactive and animated applications are human-centric, and the representation of human motion is unavoidable, making the representation of human motions ubiquitous. Animators strive to computationally process human motion in a way that the motions appear realistic in animated applications. Motion style transfer is an attractive technique that is widely used to create realistic motions in near real-time. motion style transfer approach employs existing captured motion data to generate realistic samples automatically and updates the motion data accordingly. This approach eliminates the need for handcrafted motions from scratch for every frame. The popularity of deep learning (DL) algorithms reshapes motion style transfer approaches, as such algorithms can predict subsequent motion styles. The majority of motion style transfer approaches use different variants of deep neural networks (DNNs) to accomplish motion style transfer approaches. This paper provides a comprehensive comparative analysis of existing state-of-the-art DL-based motion style transfer approaches. The enabling technologies that facilitate motion style transfer approaches are briefly presented in this paper. When employing DL-based methods for motion style transfer, the selection of the training dataset plays a key role in the performance. By anticipating this vital aspect, this paper provides a detailed summary of existing well-known motion datasets. As an outcome of the extensive overview of the domain, this paper highlights the contemporary challenges faced by motion style transfer approaches. [ABSTRACT FROM AUTHOR]

A chilled water system transports cooling functionality from refrigerators to users via heating, ventilation, and air conditioning (HVAC) systems. This paper investigated an optimal control strategy to regulate the volume flow rate of each user branch in a chilled water system, considering the minimum resistance operation to reduce energy consumption. An artificial neural network (ANN) was adopted to establish the nonlinear relationship between the volume flow rate of each user branch, pump frequency, and valve opening of each user branch. An optimal control strategy for a chilled water HVAC system is proposed in this article, according to the pump-valve combined control (PVCC) principle and an ANN model, i.e., pump-valve combined control using an artificial neural network model (PVCC-ANN). A series of tests were conducted to collect data to train the ANN model and analyze the performance of the PVCC-ANN in an experimental chilled water system. The results show that the trained ANN model has good prediction performance. A minimum resistance operation can be achieved to control the volume flow rate of each user branch independently by using the PVCC-ANN model. Moreover, the proposed PVCC-ANN method shows good energy-saving performance in chilled water systems, which can be attributed to the minimum resistance operation. Taking the energy consumption of the pump's constant frequency operation as the reference, the energy saving rate using the PVCC-ANN is between 14.3% and 58.6% under 10 operating conditions, as reported in this paper. [ABSTRACT FROM AUTHOR]

Increasing connection rates of rooftop photovoltaic (PV) systems to electricity distribution networks has become a major concern for the distribution network service providers (DNSPs) due to the inability of existing network infrastructure to accommodate high levels of PV penetration while maintaining voltage regulation and other operational requirements. The solution to this dilemma is to undertake a hosting capacity (HC) study to identify the maximum penetration limit of rooftop PV generation and take necessary actions to enhance the HC of the network. This paper presents a comprehensive review of two topics: HC assessment strategies and reinforcement learning (RL)-based coordinated voltage control schemes. In this paper, the RL-based coordinated voltage control schemes are identified as a means to enhance the HC of electricity distribution networks. RL-based algorithms have been widely used in many power system applications in recent years due to their precise, efficient and model-free decision-making capabilities. A large portion of this paper is dedicated to reviewing RL concepts and recently published literature on RL-based coordinated voltage control schemes. A non-exhaustive classification of RL algorithms for voltage control is presented and key RL parameters for the voltage control problem are identified. Furthermore, critical challenges and risk factors of adopting RL-based methods for coordinated voltage control are discussed. [ABSTRACT FROM AUTHOR]

Marine oil spills can cause serious damage to marine ecosystems and biological species, and the pollution is difficult to repair in the short term. Accurate oil type identification and oil thickness quantification are of great significance for marine oil spill emergency response and damage assessment. In recent years, hyperspectral remote sensing technology has become an effective means to monitor marine oil spills. The spectral and spatial features of oil spill images at different levels are different. To accurately identify oil spill types and quantify oil film thickness, and perform better extraction of spectral and spatial features, a multilevel spatial and spectral feature extraction network is proposed in this study. First, the graph convolutional neural network and graph attentional neural network models were used to extract spectral and spatial features in non-Euclidean space, respectively, and then the designed modules based on 2D expansion convolution, depth convolution, and point convolution were applied to extract feature information in Euclidean space; after that, a multilevel feature fusion method was developed to fuse the obtained spatial and spectral features in Euclidean space in a complementary way to obtain multilevel features. Finally, the multilevel features were fused at the feature level to obtain the oil spill information. The experimental results show that compared with CGCNN, SSRN, and A2S2KResNet algorithms, the accuracy of oil type identification and oil film thickness classification of the proposed method in this paper is improved by 12.82%, 0.06%, and 0.08% and 2.23%, 0.69%, and 0.47%, respectively, which proves that the method in this paper can effectively extract oil spill information and identify different oil spill types and different oil film thicknesses. [ABSTRACT FROM AUTHOR]

Intracranial hemorrhage (ICH) is a serious medical condition that necessitates a prompt and exhaustive medical diagnosis. This paper presents a multi-label ICH classification issue with six different types of hemorrhages, namely epidural (EPD), intraparenchymal (ITP), intraventricular (ITV), subarachnoid (SBC), subdural (SBD), and Some. A patient may experience numerous hemorrhages at the same time in some situations. A CT scan of a patient's skull is used to detect and classify the type of ICH hemorrhage(s) present. First, our model determines whether there is a hemorrhage or not; if there is a hemorrhage, the model attempts to identify the type of hemorrhage(s). In this paper, we present a hybrid deep learning approach that combines convolutional neural network (CNN) and Long-Short Term Memory (LSTM) approaches (Conv-LSTM). In addition, to propose viable solutions for the problem, we used a Systematic Windowing technique with a Conv-LSTM. To ensure the efficacy of the proposed model, experiments are conducted on the RSNA dataset. The suggested model provides higher sensitivity (93.87%), specificity (96.45%), precision (95.21%), and accuracy (95.14%). In addition, the obtained F1 score results outperform existing deep neural network-based algorithms. [ABSTRACT FROM AUTHOR]

Hyperinsulinemia is a condition characterized by excessively high levels of insulin in the bloodstream. It can exist for many years without any symptomatology. The research presented in this paper was conducted from 2019 to 2022 in cooperation with a health center in Serbia as a large cross-sectional observational study of adolescents of both genders using datasets collected from the field. Previously used analytical approaches of integrated and relevant clinical, hematological, biochemical, and other variables could not identify potential risk factors for developing hyperinsulinemia. This paper aims to present several different models using machine learning (ML) algorithms such as naive Bayes, decision tree, and random forest and compare them with a new methodology constructed based on artificial neural networks using Taguchi's orthogonal vector plans (ANN-L), a special extraction of Latin squares. Furthermore, the experimental part of this study showed that ANN-L models achieved an accuracy of 99.5% with less than seven iterations performed. Furthermore, the study provides valuable insights into the share of each risk factor contributing to the occurrence of hyperinsulinemia in adolescents, which is crucial for more precise and straightforward medical diagnoses. Preventing the risk of hyperinsulinemia in this age group is crucial for the well-being of the adolescents and society as a whole. [ABSTRACT FROM AUTHOR]

This paper investigates the effect of the architectural design of deep learning models in combination with a feature engineering approach considering the temporal variation in the features in the case of tropospheric ozone forecasting. Although deep neural network models have shown successful results by extracting features automatically from raw data, their performance in the domain of air quality forecasting is influenced by different feature analysis approaches and model architectures. This paper proposes a simple but effective analysis of tropospheric ozone time series data that can reveal temporal phases of the ozone evolution process and assist neural network models to reflect these temporal variations. We demonstrate that addressing the ozone evolution phases when developing the model architecture improves the performance of deep neural network models. As a result, we evaluated our approach on the CNN model and showed that not only does it improve the performance of the CNN model, but also that the CNN model in combination with our approach boosts the performance of the other deep neural network models such as LSTM. The development of the CNN, LSTM-CNN, and CNN-LSTM models using the proposed approach improved the prediction performance of the models by 3.58%, 1.68%, and 3.37%, respectively. [ABSTRACT FROM AUTHOR]

Simple Summary: In radiotherapy treatment, organs at risk and target volumes are contoured by the clinicians to prepare a dosimetry plan. In retrospective data, these structures are not often standardised to universal names across the patients plans, which is required to enable data mining and analysis. In this paper, a new method was proposed and evaluated to automatically standardise radiotherapy structures names using machine learning algorithms. The proposed approach was deployed over a dataset with 1613 patients collected from Liverpool & Macarthur Cancer Therapy Centres, New South Wales, Australia. It was concluded that machine learning techniques can standardise the dosimetry plan structures, taking into consideration the integration of multiple modalities representing each structure during the training process. In progressing the use of big data in health systems, standardised nomenclature is required to enable data pooling and analyses. In many radiotherapy planning systems and their data archives, target volumes (TV) and organ-at-risk (OAR) structure nomenclature has not been standardised. Machine learning (ML) has been utilised to standardise volumes nomenclature in retrospective datasets. However, only subsets of the structures have been targeted. Within this paper, we proposed a new approach for standardising all the structures nomenclature by using multi-modal artificial neural networks. A cohort consisting of 1613 breast cancer patients treated with radiotherapy was identified from Liverpool & Macarthur Cancer Therapy Centres, NSW, Australia. Four types of volume characteristics were generated to represent each target and OAR volume: textual features, geometric features, dosimetry features, and imaging data. Five datasets were created from the original cohort, the first four represented different subsets of volumes and the last one represented the whole list of volumes. For each dataset, 15 sets of combinations of features were generated to investigate the effect of using different characteristics on the standardisation performance. The best model reported 99.416% classification accuracy over the hold-out sample when used to standardise all the nomenclatures in a breast cancer radiotherapy plan into 21 classes. Our results showed that ML based automation methods can be used for standardising naming conventions in a radiotherapy plan taking into consideration the inclusion of multiple modalities to better represent each volume. [ABSTRACT FROM AUTHOR]

Radial pulse signals are produced by the periodic ejection of blood from the heart, and physiological and pathological information of the human body can be analyzed by extracting the time-domain characteristics of pulse waves. However, since pulse signals are weak physiological signals on the body surface and complex, the acquisition of pulse characteristics using the traditional curvature method will produce a large error, which cannot meet the needs of pulse wave analysis in current clinical practice. To solve this problem, a multi-morphological pulse signal feature recognition algorithm based on the one-dimensional deep convolutional neural network (1D-DCNN) model is proposed. We used the multi-channel pulse diagnosis instrument independently developed by the team to collect radial pulse signals under continuous pressure of the test subjects and collected 115 subjects and extracted a total of 1300 single-cycle pulse signals and then divided these pulse signals into 6 different forms. Five types of pulse signal time-domain feature points were labeled, and five independent feature point datasets were labeled and formed five customized neural network models that were generated to train and identify the pulse feature point datasets independently. The results show that the correction coefficient ( R a d j u s t e d 2 ) of the multi-class pulse signal processing algorithm proposed in this paper for each type of feature point recognition reaches more than 0.92. The performance is significantly better than that of the traditional curvature method, which shows the accuracy and superiority of the proposed method. Therefore, the multi-class pulse signal characteristic parameter recognition model based on the 1D-DCNN model proposed in this paper can efficiently and accurately identify pulse time-domain characteristic parameters, which can be applied to discriminate time-domain pulse information in clinical practice and assist doctors in diagnosis. [ABSTRACT FROM AUTHOR]

Sentiment classification is a key task in exploring people's opinions; improved sentiment classification can help individuals make better decisions. Social media users are increasingly using both images and text to express their opinions and share their experiences, instead of only using text in conventional social media. As a result, understanding how to fully utilize them is critical in a variety of activities, including sentiment classification. In this work, we provide a fresh multimodal sentiment classification approach: visual distillation and attention network or VisdaNet. First, this method proposes a knowledge augmentation module, which overcomes the lack of information in short text by integrating the information of image captions and short text; secondly, aimed at the information control problem in the multi-modal fusion process in the product review scene, this paper proposes a knowledge distillation based on the CLIP module to reduce the noise information of the original modalities and improve the quality of the original modal information. Finally, regarding the single-text multi-image fusion problem in the product review scene, this paper proposes visual aspect attention based on the CLIP module, which correctly models the text-image interaction relationship in special scenes and realizes feature-level fusion across modalities. The results of the experiment on the Yelp multimodal dataset reveal that our model outperforms the previous SOTA model. Furthermore, the ablation experiment results demonstrate the efficacy of various tactics in the suggested model. [ABSTRACT FROM AUTHOR]

The well-known dispatcher training simulator (DTS), as a good tool to train power system dispatchers, has been widely used for over 40 years. However, with the high-speed development of the smart grid, the traditional DTSs have struggled to meet the power industry's expectations. To enhance the effectiveness of dispatcher training, technical innovations in DTSs are becoming more and more demanding. Meanwhile, the ever-advancing artificial intelligence (AI) technology provides the basis for the design of intelligent DTSs. This paper systematically reviews the traditional DTS in terms of its origin, structure, and functions, as well as limitations in the context of the smart grid. Then, this paper summarizes the AI techniques commonly used in the field of power systems, such as expert systems, artificial neural networks, and the fuzzy set theory, and employs them to develop intelligent DTSs. Regarding a less studied aspect of DTSs, i.e., intelligent training control, we introduce the Adaptive Learning System (ALS) to develop a personalized training program, which will also be an important aspect of future research. [ABSTRACT FROM AUTHOR]

In recent years, we have been witnessing a growing interest in the subject of communication at sea. One of the promising solutions to enable widespread access to data transmission capabilities in coastal waters is the possibility of employing an on-shore wireless access infrastructure. However, such an infrastructure is a heterogeneous one, managed by many independent operators and utilizing a number of different communication technologies. If a moving sea vessel is to maintain a reliable communication within such a system, it needs to employ a set of network mechanisms dedicated for this purpose. In this paper, we provide a short overview of such requirements and overall characteristics of maritime communication, but our main focus is on the link selection procedure—an element of critical importance for the process of changing the device/system which the mobile vessel uses to retain communication with on-shore networks. The paper presents the concept of employing deep neural networks for the purpose of link selection. The proposed methods have been verified using propagation models dedicated to realistically represent the environment of maritime communications and compared to a number of currently popular solutions. The results of evaluation indicate a significant gain in both accuracy of predictions and reduction of the amount of test traffic which needs to be generated for measurements. [ABSTRACT FROM AUTHOR]

The article presents the application of a hyperspectral camera in mobile robot navigation. Hyperspectral cameras are imaging systems that can capture a wide range of electromagnetic spectra. This feature allows them to detect a broader range of colors and features than traditional cameras and to perceive the environment more accurately. Several surface types, such as mud, can be challenging to detect using an RGB camera. In our system, the hyperspectral camera is used for ground recognition (e.g., grass, bumpy road, asphalt). Traditional global path planning methods take the shortest path length as the optimization objective. We propose an improved A* algorithm to generate the collision-free path. Semantic information makes it possible to plan a feasible and safe path in a complex off-road environment, taking traveling time as the optimization objective. We presented the results of the experiments for data collected in a natural environment. An important novelty of this paper is using a modified nearest neighbor method for hyperspectral data analysis and then using the data for path planning tasks in the same work. Using the nearest neighbor method allows us to adjust the robotic system much faster than using neural networks. As our system is continuously evolving, we intend to examine the performance of the vehicle on various road surfaces, which is why we sought to create a classification system that does not require a prolonged learning process. In our paper, we aimed to demonstrate that the incorporation of a hyperspectral camera can not only enhance route planning but also aid in the determination of parameters such as speed and acceleration. [ABSTRACT FROM AUTHOR]

In the discipline of hand gesture and dynamic sign language recognition, deep learning approaches with high computational complexity and a wide range of parameters have been an extremely remarkable success. However, the implementation of sign language recognition applications for mobile phones with restricted storage and computing capacities is usually greatly constrained by those limited resources. In light of this situation, we suggest lightweight deep neural networks with advanced processing for real-time dynamic sign language recognition (DSLR). This paper presents a DSLR application to minimize the gap between hearing-impaired communities and regular society. The DSLR application was developed using two robust deep learning models, the GRU and the 1D CNN, combined with the MediaPipe framework. In this paper, the authors implement advanced processes to solve most of the DSLR problems, especially in real-time detection, e.g., differences in depth and location. The solution method consists of three main parts. First, the input dataset is preprocessed with our algorithm to standardize the number of frames. Then, the MediaPipe framework extracts hands and poses landmarks (features) to detect and locate them. Finally, the features of the models are passed after processing the unification of the depth and location of the body to recognize the DSL accurately. To accomplish this, the authors built a new American video-based sign dataset and named it DSL-46. DSL-46 contains 46 daily used signs that were presented with all the needed details and properties for recording the new dataset. The results of the experiments show that the presented solution method can recognize dynamic signs extremely fast and accurately, even in real-time detection. The DSLR reaches an accuracy of 98.8%, 99.84%, and 88.40% on the DSL-46, LSA64, and LIBRAS-BSL datasets, respectively. [ABSTRACT FROM AUTHOR]

The frequency-hopping spread spectrum (FHSS) technique is widely used in secure communications. In this technique, the signal carrier frequency hops over a large band. The conventional non-compressed receiver must sample the signal at high rates to catch the entire frequency-hopping range, which is unfeasible for wide frequency-hopping ranges. In this paper, we propose an efficient adaptive compressed method to measure and detect the FHSS signals non-cooperatively. In contrast to the literature, the FHSS signal-detection method proposed in this paper is achieved directly with compressed sampling rates. The measurement kernels (the non-zero coefficients in the measurement matrix) are designed adaptively, using continuously updated knowledge from the compressed measurement. More importantly, in contrast to the iterative optimizations of the measurement matrices in the literature, the deep neural networks are trained once using task-specific information optimization and repeatedly implemented for measurement kernel design, enabling efficient adaptive detection of the FHSS signals. Simulations show that the proposed method provides stably low missing detection rates, compared to the compressed detection with random measurement kernels and the recently proposed method. Meanwhile, the measurement design in the proposed method is shown to provide improved efficiency, compared to the commonly used recursive method. [ABSTRACT FROM AUTHOR]

With the development of data collection and storage capabilities in recent decades, abundant data have been accumulated in geotechnical engineering fields, providing opportunities for the usage of machine learning approaches. Thus, a rising number of scholars are adopting machine learning techniques to settle geotechnical issues. In this paper, the application of three popular machine learning algorithms, support vector machine (SVM), artificial neural network (ANN), and decision tree (DT), as well as other representative algorithms in geotechnical engineering, is reviewed. Meanwhile, the applicability of diverse machine learning algorithms in settling specific geotechnical engineering issues is compared. The main findings are as follows: ANN, SVM, and DT have been widely adopted to solve a variety of geotechnical engineering issues, such as the classification of soil and rock types, predicting the properties of geotechnical materials, etc. Based on the collected relevant research, the performance of random forest (RF) in sorting soil types and assessing landslide susceptibility is satisfying; SVM has high precision in classifying rock types and forecasting rock deformation; and backpropagation ANNs and Hopfield ANNs are recommended to forecast rock compressive strength and soil settlement, respectively. [ABSTRACT FROM AUTHOR]

This paper introduces a new approach to controlling Pressure Swing Adsorption (PSA) using a neural network controller based on a Model Predictive Control (MPC) process. We use a Hammerstein–Wiener (HW) model representing the real PSA process data. Then, we design an MPC-controlled model based on the HW model to maintain the bioethanol purity near 99 % molar fraction. This work proposes an Artificial Neural Network (ANN) that captures the dynamics of the PSA model controlled by the MPC strategy. Both controllers are validated using the HW model of the PSA process, showing great performance and robustness against disturbances. The results show that we can follow the desired trajectory and attenuate disturbances, achieving the purity of bioethanol at a molar fraction value of 0.99 using the ANN based on the MPC strategy with 94 % of fit in the control signal and a 97 % fit in the purity signal, so we can conclude that our ANN can be used to attenuate disturbances and maintain purity in the PSA process. [ABSTRACT FROM AUTHOR]

Recently, there has been a growth in the number of studies addressing the automatic processing of low-resource languages. The lack of speech and text data significantly hinders the development of speech technologies for such languages. This paper introduces an automatic speech recognition system for Livvi-Karelian. Acoustic models based on artificial neural networks with time delays and hidden Markov models were trained using a limited speech dataset of 3.5 h. To augment the data, pitch and speech rate perturbation, SpecAugment, and their combinations were employed. Language models based on 3-grams and neural networks were trained using written texts and transcripts. The achieved word error rate metric of 22.80% is comparable to other low-resource languages. To the best of our knowledge, this is the first speech recognition system for Livvi-Karelian. The results obtained can be of a certain significance for development of automatic speech recognition systems not only for Livvi-Karelian, but also for other low-resource languages, including the fields of speech recognition and machine translation systems. Future work includes experiments with Karelian data using techniques such as transfer learning and DNN language models. [ABSTRACT FROM AUTHOR]

Recycling aluminium is essential for a circular economy, reducing the energy required and greenhouse gas emissions compared to extraction from virgin ore. A 'Twitch' waste stream is a mix of shredded wrought and cast aluminium. Wrought must be separated before recycling to prevent contamination from the impurities present in the cast. In this paper, we demonstrate magnetic induction spectroscopy (MIS) to classify wrought from cast aluminium. MIS measures the scattering of an oscillating magnetic field to characterise a material. The conductivity difference between cast and wrought makes it a promising choice for MIS. We first show how wrought can be classified on a laboratory system with 89.66% recovery and 94.96% purity. We then implement the first industrial MIS material recovery solution for sorting Twitch, combining our sensors with a commercial-scale separator system. The industrial system did not reflect the laboratory results. The analysis found three areas of reduced performance: (1) metal pieces correctly classified by one sensor were misclassified by adjacent sensors that only captured part of the metal; (2) the metal surface facing the sensor can produce different classification results; and (3) the choice of machine learning algorithm is significant with artificial neural networks producing the best results on unseen data. [ABSTRACT FROM AUTHOR]

Multiconnectivity allows user equipment/devices to connect to multiple radio access technologies simultaneously, including 5G, 4G (LTE), and WiFi. It is a necessity in meeting the increasing demand for mobile network services for the 5G and beyond wireless networks, while ensuring that mobile operators can still reap the benefits of their present investments. Multipath TCP (MPTCP) has been introduced to allow uninterrupted reliable data transmission over multiconnectivity links. However, energy consumption is a significant issue for multihomed wireless devices since most of them are battery-powered. This paper employs software-defined networking (SDN) and deep neural networks (DNNs) to manage the energy consumption of devices with multiconnectivity running MPTCP. The proposed method involves two lightweight algorithms implemented on an SDN controller, using a real hardware testbed of dual-homed wireless nodes connected to WiFi and cellular networks. The first algorithm determines whether a node should connect to a specific network or both networks. The second algorithm improves the selection made by the first by using a DNN trained on different scenarios, such as various network sizes and MPTCP congestion control algorithms. The results of our extensive experimentation show that this approach effectively reduces energy consumption while providing better network throughput performance compared to using single-path TCP or MPTCP Cubic or BALIA for all nodes. [ABSTRACT FROM AUTHOR]