Most research in human activity recognition is supervised, while non-supervised approaches are not completely unsupervised. Moreover, These methods cannot be used in real-time applications due to high calculations. In this paper, we provide a novel flexible multi-objective particle swarm optimization clustering method based on game theory (FMOPG) to discover human activities fully unsupervised. Unlike conventional clustering methods that estimate the number of clusters and are very time-consuming and inaccurate, an incremental technique is introduced which makes the proposed method flexible in dealing with the number of clusters and improves the speed of clustering. By adopting this technique, clusters with a better connectedness and good separation from other clusters are gradually selected. Updating of particles’ velocity is modified by adopting the concept of mean-shift vector to improve the convergence speed of PSO in achieving the best solution and dealing with non-spherical shape clusters. Multi-objective optimization problem is mapped to game theory by adopting Nash equilibrium to select the optimal solution on the pareto front. Gaussian mutation is also employed on the pareto front to generate diverse solutions and create a balance between exploitation and exploration. Moreover, A smart grid-based method is proposed to initialize the population to generate diverse solutions and reduce the variance between the worst and best clustering results. The proposed method is compared with state-of-the-art methods on seven challenging datasets. FMOPG has improved clustering accuracy by 3.65 % compared to other automated methods. Moreover, the incremental technique has improved the clustering time by 71.18 %.

This study presents an algorithm for constructing a product phylogenetic tree that can observe product evolutionary patterns. This study was based on previous studies of modeling technology and science evolution that used natural language processing techniques and network analysis, and a product phylogenetic tree construction algorithm that can clearly reflect the complexity of products and the ancestor-descendant relationship between products was developed. To verify the proposed algorithm, a mobile product phylogenetic tree was empirically constructed using 9,803 mobile product feature data released from 1995 to 2019, and smartphone product speciation, feature phone and pseudo-smartphone extinction, and a Chinese mobile product rise in the late 2010s were observed through the mobile product phylogenetic tree. In addition, it was possible to observe the behaviors of firms that produced products through a product phylogenetic tree: the existence of Apple in the mainstream of mobile products since 2007, the catch-up strategy of Samsung, the extinction of Nokia due to incorrect operating system Symbian adoption, and the strategic failure of LG. The fact that the behaviors of products and firms observed through the mobile product phylogenetic tree explain the actual events well demonstrate that the proposed product phylogenetic tree construction algorithm is appropriate. In conclusion, this study contributes to the improvement of the previous product phylogenetic tree construction methodology by adapting natural language processing and network analysis. It is expected that the product phylogenetic tree construction algorithm in this research can be established as a basic approach for product evolution and product innovation research.

Process industry is the pillar industry of national economy, particularly, the process of producing magnesia by fused magnesia furnace system is a typical category of process industry. Due to the complex smelting mechanism and changing production factors, abnormal working conditions often occur in fused magnesia furnace. The semi-molten condition is the most typical and harmful abnormal condition. In this paper, an adaptive pretraining-inference-dynamic training-validation semantic segmentation method based on industrial video is proposed for dynamic prediction of semi-molten condition of multiple fused magnesium furnaces. The experimental results show that compared with the prediction model without adaptive learning, the prediction performance of the adaptive learning model in this paper for multiple fused magnesium melting processes is significantly improved.

Auditory perception of emotions in speech is relevant for humans to optimally navigate the social environment. While sensory perception is known to be influenced by bodily internal states such as anxiety and ambient noise, their relationship on human auditory perception is relatively less understood. In a supervised, internet-based experiment carried out sans the artificially controlled laboratory environment, we asked if the detection sensitivity of emotions conveyed by human speech-in-noise (acoustic signals) is modulated by individual differences in affective internal states, e.g., anxiety. In a task, participants (n=24) accurately discriminated the target emotion conveyed by the temporally unpredictable acoustic signals (signal to noise ratio=10dB), which were manipulated at four levels (Happy, Neutral, Fear and Disgust). We calculated empirical area under the curve (measure of acoustic signal detection sensitivity) based on signal detection theory to quantify our results. Specifically, Disgust and Fear detection sensitivities worsened with increasing individual severities of trait-anxiety. Further, a similar effect was evident when averaging across all four emotions. Altogether, the results suggest that individual trait-anxiety levels moderate the detection of emotions from speech-in-noise, especially those conveying negative/threatening affect. The findings may be relevant for expanding the understanding pertaining to auditory perception anomalies underlying affective states and disorders

In this paper, we present Flare: a lightweight plug & play machine learning based Event and intrusion detection systems (EIDS), which can effectively learn to discriminate among transmission line faults, contingencies, false data injection (FDI) and false command injection (FCI) attacks

A lot of progress has been made in the field of augmented reality tracking over the last few years. Several dependable tracking systems that provide real-time tracking in an indoor setting have been developed. As previously stated in this survey report, there are still some challenges with AR tracking, such as occlusion, varying lighting environments, variable backdrop contrast, and varying color intensities. The development of algorithms that use marker-based tracking to enable robust and accurate tracking in a variety of illumination, contrast, and lighting conditions will be a future focus.

The Metaverse is a virtual environment where users are represented by avatars to navigate a virtual world, which has strong links with the physical one. State-of-the-art Metaverse architectures rely on a cloud-based approach for avatar physics emulation and graphics rendering computation. Such centralized design is unfavorable as it suffers from several drawbacks caused by the long latency required for cloud access, such as low quality visualization. To solve this issue, in this paper, we propose a Fog-Edge hybrid computing architecture for Metaverse applications that leverage an edge-enabled distributed computing paradigm, which makes use of edge devices computing power to fulfil the required computational cost for heavy tasks such as collision detection in virtual universe and computation of 3D physics in virtual simulation. The computational cost related to an entity in the Metaverse such as collision detection or physics emulation are performed at the end-device of the associated physical entity. To prove the effectiveness of the proposed architecture, we simulate a distributed social metaverse application. Simulation results shows that the proposed architecture can reduce the latency by 50\% when compared with the legacy cloud-based Metaverse applications.

The increasing availability of quantum computers motivates researching their potential capabilities in enhancing the performance of data analysis algorithms. Similarly as in other research communities, also in Remote Sensing (RS) it is not yet defined how its applications can benefit from the usage of quantum computing. This paper proposes a formulation of the Support Vector Regression (SVR) algorithm that can be executed by D-Wave quantum computers. Specifically, the SVR is mapped to a Quadratic Unconstrained Binary Optimization (QUBO) optimization problem that is solved with Quantum Annealing (QA). The algorithm is tested on two different types of computing environments offered by D-Wave: The Advantage system, which directly embeds the problem into the Quantum Processing Unit (QPU), and a Hybrid solver that employs both classical and quantum computing resources. For the evaluation, we considered a biophysical variable estimation problem with RS data. The experimental results show that the proposed quantum SVR implementation can achieve comparable or in some cases better results than the classical implementation. This work is one of the first attempts to provide insight into how QA could be exploited and integrated in future RS workflows based on Machine Learning algorithms.

Unsupervised neural networks (NNs) specialize in mining potential patterns from unlabeled data in a self-organizing manner. Recently, they have also been employed as observers for process monitoring using the generated residual signals. However, few studies have explained the model behavior and analyzed the monitoring performance of unsupervised NNs awing to their multilayer nonlinear structure. The interpretability of NN covers the analysis of its working principle to seek better network design and achieve improved performance. Thus, this paper develops explainable residual generators based on unsupervised NNs, which are applicable to deep autoencoder (DAE) and variational autoencoder (VAE). Through Taylor expansion, the residual deviation caused by the fault signal, a.k.a. the fault-affected term, is proven not to disappear in the presence of a non-zero Hessian matrix. Then, the consistency of achieving the optimal monitoring performance and the training loss of NNs is presented. A new indicator function is established based on the sum martingale, a representative of a weakly dependent stochastic process. Freedman’s inequality is first applied to describe the reliability of the learned thresholds during fault evaluation, which requires a smaller sample size for training. Finally, simulations on the continuous stirred tank reactor verify the effectiveness of the proposed methods.

In this research, we present a model architecture for employing reinforcement learning with transformers for multimodal tasks. Using transformers allows us to take advantage of the transformer architecture’s simplicity and scalability, as well as developments in language and vision modelling such as ViT, GPT-x, and BERT. Specifically we have trained the model to recognize various digits from the MNIST dataset from their associated word labels and instructions provided. The approach is similar to how an infant would learn to associate pictorial representation of digits to their corresponding words. We have used a MoME transformer in conjunction with Deep Q Learning to train our model. The image inputs have been embedded using pre-trained ResNet18 and the instructions have been embedded using GLoVe before passing them to the model for prediction and training.

Sensory data is essential for the training of methods in autonomous driving like object detection, odometry, or SLAM. MEMS LiDAR sensors can be very valuable for autonomous vehicles because they are less prone to shock and wear compared to motorized optomechanical LiDAR sensors. Recording real-world data is complicated and expensive. An alternative is simulated data, but for MEMS LiDAR sensors there is no publicly available software to simulate this type of sensor. With this paper, we introduce a method to simulate data recorded by a MEMS LiDAR sensor like the Blickfeld Cube~1 (and other MEMS LiDAR sensors as well). For this, we use the open-source autonomous driving simulation environment CARLA (our method is available online\footnote[1]{https://github.com/BerensRWU/MEMS-LiDAR-Generator}). We compare our synthetic point cloud with a real-world point cloud and evaluate the similarity. Moreover, we demonstrate the application of our method on the problem of the optimal sensor configuration.

A new subclass of Sallen-Key RC-filters with independent adjustment of the main parameters, which are implemented on voltage followers (buffer and operational amplifiers), has been studied. Low- and high-pass filters, band-pass and notch filters are offered, which have the ability to independently adjust the pole frequency, pole quality factor and transmission coefficient with different resistors. Their computer simulation was carried out in the Micro-Cap environment, the amplitude-frequency and phase-frequency characteristics were obtained in the tuning mode using the models of buffer and operational amplifiers built into the Micro-Cap.

This paper investigates the performance of non-orthogonal multiple access (NOMA) based hybrid satellite-unmanned aerial vehicle (UAV) systems, where a low Earth orbit (LEO) satellite communicates with the ground users via a decode and forward (DF) UAV relay. We investigate a two NOMA users system, where a far user (FU) and a near user (NU) are served by the UAV which is located at a certain height above the origin of the coverage circle. The channel between satellite and UAV is assumed to follow a Shadowed-Rician fading and the channels between UAV and users are assumed to follow a Nakagami-m fading. New closed-form expressions of the outage probabilities for the two users and the system are derived. Different from other work in literature, we take into consideration different parameters affecting the total link budget. Additionally, we propose an algorithm for minimizing the system outage probability. The mathematical analysis is verified by extensive representative Monte-Carlo (MC) simulations. Finally, simulations are provided to demonstrate the impact of important parameters on the considered system as well as the superiority of the NOMA scheme the over reference scheme.

he main problem that we face today is not about getting services, but how efficiently we manage them. We are living in an era where we cannot neglect the fact that we are now depending on various electronic devices which we use daily in our life. These devices have become an inseparable part of our life, in the present scenario technology is evolving so rapidly that it’s a fact if we do not accept any upcoming trend then we lag behind in the eye of others. Managing all these devices and remembering to tune them is quite a hectic task, as we have to do other main tasks too. Using every device efficiently at the same time is not possible. Hence, we came up with a solution, with a system that will manage all these devices according to predefined rules or instructions. Current smart home systems work in a way, to establish a connection between nodes and a hub and are controlled by smartphones, computers, remote, etc. In this research paper, we have proposed a model that will not only establish a connection but will work smartly according to instructions and algorithms saved earlier. This Auxiliary Informatics System (AuxIS) will serve the requirement of managing all the devices in a parallel fashion with very little human intervention and hence, will make a smarter home environment.

The Internet of Vehicles over Vehicular Ad-hoc Networks is an emerging technology enabling the development of smart city applications focused on improving traffic safety, traffic efficiency, and the overall driving experience. These applications have stringent requirements detailed in Service Level Agreement. Since vehicles have limited computational and storage capabilities, applications requests are offloaded onto an integrated edge-cloud computing system. Existing offloading solutions focus on optimizing the application’s Quality of Service (QoS) in terms of execution time, and respecting a single SLA constraint. They do not consider the impact of overlapped multi-requests processing nor the vehicle’s varying speed. This paper proposes a novel Artificial Intelligence QoS-SLA-aware adaptive genetic algorithm (QoS-SLA-AGA) to optimize the application’s execution time for multi-request offloading in a heterogeneous edge-cloud computing system, which considers the impact of processing multi-requests overlapping and dynamic vehicle speed. The proposed genetic algorithm integrates an adaptive penalty function to assimilate the SLA constraints regarding latency, processing time, deadline, CPU, and memory requirements. Numerical experiments and analysis compare our QoS-SLA-AGA to random offloading, and baseline genetic-based approaches. Results show QoS-SLA-AGA executes the requests 1.22 times faster on average compared to the random offloading approach and with 59.9% fewer SLA violations. In contrast, the baseline genetic-based approach increases the requests’ performance by 1.14 times, with 19.8% more SLA violations.

Our manuscript aims to tackle the two tricky problems in pose estimation, confusion caused by the complex background interferences and small persons with low-resolution. Those two problems, even the state-of-the-art methods, such as MultiPoseNet, Detectron2, can’t solve them well. But we achieve superior performances than them. Our method surpasses the state-of-the-art methods because our novel auxiliary network, which has never appeared in all the previous networks.

Background: The quick detection of needs and resources during a disaster can save lives. Twitter is a reliable information source during disasters and has been studied using machine learning for situational awareness. There are many methods to utilize machine learning to detect needs and resource availability via Twitter, but the most common and accurate methods remain unclear. This scoping review addresses this gap. Methods: Keywords were defined within the concepts of machine learning, disasters, and classifying needs and resources. After the database searches, PRISMA guidelines were followed to perform a partnered, two-round scoping literature review. Results: 42 articles met the inclusion criteria for analysis. Geographically, the largest portion of the studies took place in the United States (24.5%), followed by Nepal (18.4%), and India (16.3%). The most studied disaster type was earthquake (25.6%), followed by hurricane (16.7%). While there was no consensus on best methods, the most used algorithms included neural networks using different types of word embeddings to optimize performance. None of the tools were ready to be used directly by aid organizations or policymakers. Conclusion: Machine learning tools for resource allocation are needed to provide timely assistance to those in need during disasters. This review indicates a need for additional research regarding a consensus on best practices for algorithm model selection, benchmarking datasets, crisis lexicons, word embedding techniques and evaluation methods. These tools have a high potential for improving real-time emergency management across all disaster phases, especially as disasters of all kinds become more and more frequent.

Adversarial attack is to craft tiny perturbations on inputs, causing neural networks to give incorrect outputs with high confidence, while adversarial training is the de facto most successful method to obtain robust neural networks. Nevertheless, adversarial training suffers from robust overfitting: the increasing robustness gap between train and test datasets. This paper aims at reducing this robust overfitting (i.e., improving adversarially robust generalization / generalized adversarial robustness). Firstly, we study the robust bias-variance trade-off and find that the robust overfitting even happens on the simple dataset (100MNIST), which is considered free from that in previous research. Next, based on 100MNIST and Gaussian data, the correlation between feature augmentation and robust overfitting is analyzed on a theoretical basis. A novel insight is obtained: the augmentation on robust features can improve generalized adversarial robustness, through alleviating the over-optimization on non-robust features. Then, we propose a regularization term, Feature Alignment (FA), to guide synthesized sample directions. Adversarial FA aided Generative Adversarial Network, AFAGAN, generates samples aligning robust features, which can train more adversarially robust generalized models. Experiments on CIFAR10 show that augmented data of AFAGAN significantly improves the generalized adversarial robustness.