Ernando P. Batista, Alex Santos, Maycon Peixoto
et al.
The growing demand for IoT applications in highly dynamic environments with multiple connected devices introduces significant scalability and low-latency challenges. In the context of software-defined networking (SDN) integrated with Edge environments, adopting machine learning (ML) techniques has emerged as a promising approach to meet these requirements. This study presents a Systematic Literature Review (SLR) that identifies and analyzes ML-based solutions applied to Software-Defined Internet of Things (SD-IoT) infrastructures in Edge environments, emphasizing improving low latency and scalability. Following established methodological best practices, we conducted the review, including a clear definition of research questions, well-defined inclusion and exclusion criteria, a structured search protocol, and multiple scientific databases. Based on the analysis of selected studies, the main strategies employed to enhance network performance are categorized, along with the level of fidelity and complexity of the experimental environments used, and the realism and applicability of the proposed solutions are discussed. Furthermore, drawing from the context of the selected studies, the most recurrent ML approaches are presented—including supervised, unsupervised, and reinforcement learning methods—along with a discussion of their advantages and limitations in dynamic network scenarios. By compiling and organizing the contributions from the literature, this paper provides a comprehensive overview of the state of the art in applying ML to SD-IoT networks, shedding light on current trends, existing gaps, and research opportunities aimed at building more intelligent and adaptable solutions for IoT environments.
Jiri Konecny, Jaromir Konecny, Kamil Bancik
et al.
Internet of Things (IoT) sensors designed for environmental and agricultural purposes can offer significant contributions to creating a sustainable and green environment. However, powering these sensors remains a challenge, and exploiting the temperature difference between air and soil appears to be a promising solution. For energy-harvesting technologies, accurate soil temperature profile data are needed. This study uses meteorological and soil temperature profile data collected in the Czech Republic to train machine learning models based on Polynomial Regression (PR), Support Vector Regression (SVR), and Long Short-Term Memory (LSTM) to predict the soil temperature profile. The results of the study indicate an error of 0.79 °C, which is approximately 10.9% lower than the temperature error reported in state-of-the-art studies. Beyond achieving a lower temperature prediction error, the proposed solution simplifies the input parameters of the model to only ambient temperature and solar irradiance. This improvement significantly reduces the computational costs associated with the regression model, offering a more efficient approach to predicting soil temperature for the purpose of optimizing energy harvesting in IoT sensors.
For a weakly guiding, single-mode, graded-index circular optical fiber, the general form of the dependence of the propagation constant on the waveguide parameter is obtained. From Maxwell's equations, an equation for the field in a light guide with a gradient refractive index profile is derived. Using a power-law refractive index profile for the first three powers and a Gaussian index profile as examples, dependences of the propagation constant, phase and group velocities on the waveguide parameter are obtained. For the ratio of the power transferred by the mode to the total stored energy per unit length of the waveguide, a dependence on the waveguide parameter is plotted. It is shown that as the waveguide parameter increases and the degree of the power-law profile increases, the fraction of transferred power decreases and approaches the fraction of transmitted power for the Gaussian profile. The results obtained can be used to create waveguides for specific applications.
The paper addresses the problem of transforming multi-input multi-output discrete-time nonlinear state equations into the classical observer form using state transformation. Necessary and sufficient geometric solvability conditions are given in terms of vector fields. The results obtained generalize the previous ones in several aspects. First, the results are also applicable to non-reversible systems. Second, they hold almost everywhere, not only around the equilibrium point of the system. The generalizations are possible due to the use of different mathematical tools. The proof of sufficiency also provides a method for finding the state transformation. The results are illustrated by two examples.
The war in Ukraine has caused significant losses to the Ukrainian agricultural sector and threatened global food security. This study aims to comprehensively analyse the impact of the Russian invasion on Ukrainian agriculture in its initial phase, examine the responses of Ukrainian agricultural policy and outline key elements for the post-war development of the Ukrainian agricultural sector. Using a systematic approach that includes thorough data collection and analysis from official sources, the Ukrainian press, relevant legislation and statistical data, our study focuses on the first five months of the conflict, from February 24, 2022 to July 24, 2022, to identify the main challenges during this period. The agricultural policy analysis shows that the Ukrainian government has adopted a reactive approach, including tax simplification, affordable credit, deregulation, financial support for the agricultural sector, reduced prices for inputs and resources to support farmers, and improvements in logistics. Critical factors identified for post-war development include repatriation and agricultural education, support for small and medium-sized farms, integration into global markets, emphasis on organic practices and sustainable development, and digitalization in agriculture. The integration of a systematic overview of the Ukrainian agricultural sector and an analysis of key elements of post-war development provides essential insights for policy-makers and researchers concerned with the impact of war on the agricultural sector and food security.
Recently, text-to-image (T2I) synthesis has undergone significant advancements, particularly with the emergence of Large Language Models (LLM) and their enhancement in Large Vision Models (LVM), greatly enhancing the instruction-following capabilities of traditional T2I models. Nevertheless, previous methods focus on improving generation quality but introduce unsafe factors into prompts. We explore that appending specific camera descriptions to prompts can enhance safety performance. Consequently, we propose a simple and safe prompt engineering method (SSP) to improve image generation quality by providing optimal camera descriptions. Specifically, we create a dataset from multi-datasets as original prompts. To select the optimal camera, we design an optimal camera matching approach and implement a classifier for original prompts capable of automatically matching. Appending camera descriptions to original prompts generates optimized prompts for further LVM image generation. Experiments demonstrate that SSP improves semantic consistency by an average of 16% compared to others and safety metrics by 48.9%.
Neural architecture search (NAS) has emerged as a powerful paradigm that enables researchers to automatically explore vast search spaces and discover efficient neural networks. However, NAS suffers from a critical bottleneck, i.e. the evaluation of numerous architectures during the search process demands substantial computing resources and time. In order to improve the efficiency of NAS, a series of methods have been proposed to reduce the evaluation time of neural architectures. However, they are not efficient enough and still only focus on the accuracy of architectures. Beyond classification accuracy, real-world applications increasingly demand more efficient and compact network architectures that balance multiple performance criteria. To address these challenges, we propose the SMEMNAS, a pairwise comparison relation-assisted multi-objective evolutionary algorithm based on a multi-population mechanism. In the SMEMNAS, a surrogate model is constructed based on pairwise comparison relations to predict the accuracy ranking of architectures, rather than the absolute accuracy. Moreover, two populations cooperate with each other in the search process, i.e. a main population that guides the evolutionary process, while a vice population that enhances search diversity. Our method aims to discover high-performance models that simultaneously optimize multiple objectives. We conduct comprehensive experiments on CIFAR-10, CIFAR-100 and ImageNet datasets to validate the effectiveness of our approach. With only a single GPU searching for 0.17 days, competitive architectures can be found by SMEMNAS which achieves 78.91% accuracy with the MAdds of 570M on the ImageNet. This work makes a significant advancement in the field of NAS.
José Antonio Villacorta Atienza, Carlos Calvo Tapia, Valeriy A. Makarov Slizneva
The ultimate navigation efficiency of mobile robots in human environments will depend on how we will appraise them: merely as impersonal machines or as human-like agents. In the latter case, an agent may take advantage of the cooperative collision avoidance, given that it possesses recursive cognition, i.e.,the agent's decisions depend on the decisions made by humans that in turn depend on the agent's decisions. To deal with this high-level cognitive skill, we propose a neural network architecture implementing Prediction-for-CompAction paradigm. The network predicts possible human-agent collisions and compacts the time dimension by projecting a given dynamic situation into a static map. Thereby emerging compact cognitive map can be readily used as a "dynamic GPS" for planning actions or mental evaluation of the convenience of cooperation in a given context. We provide numerical evidence that cooperation yields additional room for more efficient navigation in cluttered pedestrian flows, and the agent can choose path to the target significantly shorter than a robot treated by humans as a functional machine. Moreover, the navigation safety, i.e., the chances to avoid accidental collisions, increases under cooperation. Remarkably, these benefits yield no additional load to the mean society effort. Thus, the proposed strategy is socially compliant, and the humanoid agent can behave as "one of us".
Tandem mass spectrometry (MS/MS) is a widely used technology for identifying metabolites. De novo metabolite identification is an identification strategy that does not refer to any spectral or metabolite database. However, this strategy is time-consuming and cannot meet the need for high-throughput metabolite identification. BÖcker et al. converted the de novo identification problem into the maximum colorful subtree (MCS) problem. Unfortunately, the MCS problem is NPhard, which indicates there are no existing efficient exact algorithms. To address this issue, we propose to apply quantum computing to accelerate metabolite identification. Quantum computing performs computations on quantum computers. The recent progress in this area has brought the hope of making some computationally intractable areas trackable, although there are still no general approaches to converting regular computer algorithms into quantum algorithms. Specifically, there is no efficient quantum algorithm for the MCS problem. The MCS problem can be considered as the combination of many maximum spanning tree problems that can be converted into minimum spanning tree problems. This work applies a quantum algorithm designed for the minimum spanning problem to speed up de novo metabolite identification. The possible strategy for further improving the performance is also briefly discussed.
Paulo R. G. Cordeiro, George D. C. Cavalcanti, Rafael M. O. Cruz
Dynamic Ensemble Selection (DES) is a Multiple Classifier Systems (MCS) approach that aims to select an ensemble for each query sample during the selection phase. Even with the proposal of several DES approaches, no particular DES technique is the best choice for different problems. Thus, we hypothesize that selecting the best DES approach per query instance can lead to better accuracy. To evaluate this idea, we introduce the Post-Selection Dynamic Ensemble Selection (PS-DES) approach, a post-selection scheme that evaluates ensembles selected by several DES techniques using different metrics. Experimental results show that using accuracy as a metric to select the ensembles, PS-DES performs better than individual DES techniques. PS-DES source code is available in a GitHub repository
Manual assembly workers face increasing complexity in their work. Human-centered assistance systems could help, but object recognition as an enabling technology hinders sophisticated human-centered design of these systems. At the same time, activity recognition based on hand poses suffers from poor pose estimation in complex usage scenarios, such as wearing gloves. This paper presents a self-supervised pipeline for adapting hand pose estimation to specific use cases with minimal human interaction. This enables cheap and robust hand posebased activity recognition. The pipeline consists of a general machine learning model for hand pose estimation trained on a generalized dataset, spatial and temporal filtering to account for anatomical constraints of the hand, and a retraining step to improve the model. Different parameter combinations are evaluated on a publicly available and annotated dataset. The best parameter and model combination is then applied to unlabelled videos from a manual assembly scenario. The effectiveness of the pipeline is demonstrated by training an activity recognition as a downstream task in the manual assembly scenario.
This paper expands the model for higher-order moments (evolving into an MVSK model with skewness and kurtosis analysis) and compares it to the classic quadratic objective function of Markowitz. Along with the MVSK analysis, we add an information entropy variable to the model taking into account the asset's informational efficiency and diversity, and trying to encompass the high uncertainty intrinsic to the market's returns and increase the model's validity. We analyze the practical effectiveness and the complexity of creating such a multi-objective portfolio model to see if we can provide more information to the investor with the new framework.
Mikhail Belov, Stanislav Grishko, Mikhail Lishilin
et al.
The article formulates the key aspects of the strategy for training highly qualified and in-demand professionals in the field of processing, storing and analyzing big data for a quick and high-quality solution to urgent problems of digital transformation and data-based management. The article analyzes the problems of modern full-time and distance learning and ways to overcome them using the innovative training data center "Virtual Computer Laboratory" (VCL), which incorporates the principles of self-organization and cybernetics 2.0. The author's team's view of the current requirements for basic knowledge and their harmonization with the professional standards of the APKIT association is presented, key development drivers are formulated, and potential risks are identified based on personal experience and in-depth analysis of the problems of modern IT education in the Russian Federation. The concept of a specialized training data center VCL is described, which made it possible to form a homogeneous educational environment with elements of cognitive representation of internal operational resources based on visual models and partial automation of basic technological operations using an expert system. The article also lists the key educational tasks focused on the formation of relevant core competencies, the successful solution of which became possible thanks to the developed methodology for using the VCL and its functional blocks. The possibilities of transition from traditional educational approaches to a new system-activity educational paradigm with the use of the "Virtual Computer Laboratory" data center are described.
Feras Albardi, H M Dipu Kabir, Md Mahbub Islam Bhuiyan
et al.
This study aims to explore different pre-trained models offered in the Torchvision package which is available in the PyTorch library. And investigate their effectiveness on fine-grained images classification. Transfer Learning is an effective method of achieving extremely good performance with insufficient training data. In many real-world situations, people cannot collect sufficient data required to train a deep neural network model efficiently. Transfer Learning models are pre-trained on a large data set, and can bring a good performance on smaller datasets with significantly lower training time. Torchvision package offers us many models to apply the Transfer Learning on smaller datasets. Therefore, researchers may need a guideline for the selection of a good model. We investigate Torchvision pre-trained models on four different data sets: 10 Monkey Species, 225 Bird Species, Fruits 360, and Oxford 102 Flowers. These data sets have images of different resolutions, class numbers, and different achievable accuracies. We also apply their usual fully-connected layer and the Spinal fully-connected layer to investigate the effectiveness of SpinalNet. The Spinal fully-connected layer brings better performance in most situations. We apply the same augmentation for different models for the same data set for a fair comparison. This paper may help future Computer Vision researchers in choosing a proper Transfer Learning model.
Jun-ichi Takeshita, Akinobu Toyoda, Hidenori Tani
et al.
Toxicity evaluation of chemical compounds has traditionally relied on animal experiments;however, the demand for non-animal-based prediction methods for toxicology of compounds is increasing worldwide. Our aim was to provide a classification method for compounds based on \textit{in vitro} gene expression profiles. The \textit{in vitro} gene expression data analyzed in the present study was obtained from our previous study. The data concerned nine compounds typically employed in chemical management.We used agglomerative hierarchical clustering to classify the compounds;however, there was a statistical difficulty to be overcome.We needed to properly extract RNAs for clustering from more than 30,000 RNAs. In order to overcome this difficulty, we introduced a combinatorial optimization problem with respect to both gene expression levels and the correlation between gene expression profiles. Then, the simulated annealing algorithm was used to obtain a good solution for the problem. As a result, the nine compounds were divided into two groups using 1,000 extracted RNAs. Our proposed methodology enables read-across, one of the frameworks for predicting toxicology, based on \textit{in vitro} gene expression profiles.
Using <em>Jet stream's </em>precursors, seismic locations are identified. Our research indicates that an interruption of the velocity flow lines occurs just above the epicenter approximately 3 months prior to <em>Earthquake </em>events. The duration of this phenomenon is approximately 6 – 12 hours. The average distance between epicenters and <em>Jet stream's </em>precursors is about 100 km. We explain these relationships while reviewing 8 successful <em>Earthquake </em>forecasts recently. For example: M8.3 Chile EQ on 2015/09/16; M6.6 Taiwan EQ on 2016/02/05; M7.0 Kumamoto, Japan EQ on 2016/04/15; M6.2 Italy EQ on 2016/08/24; M7.1Alaska EQ on 2018/11/30; M6.7 Chile EQ on 2019/01/20; M6.3 Japan EQ on 2019/01/08; M7.1 LA EQ on 2019/07/06. According to the hypothesis of Lithosphere-Atmosphere- <em>Ionosphere </em>Coupling (LAIC), when the <em>Jet streams </em>pass over the active epicenter region, the faults release radioactive material (ionized gases) to the atmosphere, causing a series of physical and chemical reactions, resulting in temperature and pressure changes in the atmosphere, <em>Jet streams</em>, and electric field effects in the ionosphere. A <em>Solar Induction </em>mechanism affecting the Eastern and Western Pacific Rims where most of the <em>Earthquakes </em>were successfully forecast is explored in electrical terms with a proposed <em>Plasma Tectonics </em>model.