Search Results (36)

The application of diffusion models in the field of remote sensing image generation has significantly improved the performance of generation algorithms. However, existing methods still exhibit certain limitations, such as the inability to generate images with rich texture details and minimal geometric distortions in a controllable manner. To address these shortcomings, this work introduces an innovative remote sensing image generation algorithm, OP-Gen, which is guided by textual descriptions and OpenStreetMap (OSM) images. OP-Gen incorporates two information extraction branches: ControlNet and OSM-prompt (OP). The ControlNet branch extracts structural and spatial information from OSM images and injects this information into the diffusion model, providing guidance for the overall structural framework of the generated images. In the OP branch, we design an OP-Controller module, which extracts detailed semantic information from textual prompts based on the structural information of the OSM image. This information is subsequently injected into the diffusion model, enriching the generated images with fine-grained details, aligning the generated details with the structural framework, and thus significantly enhancing the realism of the output. The proposed OP-Gen algorithm achieves state-of-the-art performance in both qualitative and quantitative evaluations. The qualitative results demonstrate that OP-Gen outperforms existing methods in terms of structural coherence and texture detail richness. Quantitatively, the algorithm achieves a Fréchet inception distance (FID) of 45.01, a structural similarity index measure (SSIM) of 0.1904, and a Contrastive Language-Image Pretraining (CLIP) score of 0.3071, all of which represent the best performance among the current algorithms of the same type. Full article

Few-shot learning has demonstrated remarkable performance in medical image segmentation. However, existing few-shot medical image segmentation (FSMIS) models often struggle to fully utilize query image information, leading to prototype bias and limited generalization ability. To address these issues, we propose the dual-filter cross attention and onion pooling network (DCOP-Net) for FSMIS. DCOP-Net consists of a prototype learning stage and a segmentation stage. During the prototype learning stage, we introduce a dual-filter cross attention (DFCA) module to avoid entanglement between query background features and support foreground features, effectively integrating query foreground features into support prototypes. Additionally, we design an onion pooling (OP) module that combines eroding mask operations with masked average pooling to generate multiple prototypes, preserving contextual information and mitigating prototype bias. In the segmentation stage, we present a parallel threshold perception (PTP) module to generate robust thresholds for foreground and background differentiation and a query self-reference regularization (QSR) strategy to enhance model accuracy and consistency. Extensive experiments on three publicly available medical image datasets demonstrate that DCOP-Net outperforms state-of-the-art methods, exhibiting superior segmentation and generalization capabilities. Full article

The remote sensing community benefits from new sensors and easier access to Earth Observation data to frequently released new land-cover maps. The propagation of such independent and heterogeneous products offers promising perspectives for various scientific domains and for the implementation and monitoring of land-use policies. Yet, it may also confuse the end-users when it comes to identifying the most appropriate product to address their requirements. Data fusion methods can help to combine competing and/or complementary maps in order to capitalize on their strengths while overcoming their limitations. We assessed the potential of the Dempster–Shafer Theory (DST) to enhance oil palm mapping in Sumatra (Indonesia) by combining four land-cover maps, hereafter named DESCALS, IIASA, XU, and MAPBIOMAS, according to the first author’s name or the research group that published it. The application of DST relied on four steps: (1) a discernment framework, (2) the assignment of mass functions, (3) the DST fusion rule, and (4) the DST decision rule. Our results showed that the DST decision map achieved significantly higher accuracy (Kappa = 0.78) than the most accurate input product (Kappa = 0.724). The best result was reached by considering the probabilities of pixels to belong to the OP class associated with DESCALS map. In addition, the belief (i.e., confidence) and conflict (i.e., uncertainty) maps produced by DST evidenced that industrial plantations were detected with higher confidence than smallholder plantations. Consequently, Kappa values computed locally were lower in areas dominated by smallholder plantations. Combining land-use products with DST contributes to producing state-of-the-art maps and continuous information for enhanced land-cover analysis. Full article

A significant challenge in utilizing hydrogen in conventional internal combustion engines is achieving a balance between NOx emissions and brake power output. A lean premixed charge (Lambda ≈ 2.5) allows for efficient and stable combustion with minimal NOx emissions. However, this comes at the cost of reduced power density due to the higher air requirements of the thermodynamic process. While supercharging can mitigate this drawback, it introduces increased complexity, cost, and size. An intriguing alternative is the 2-stroke cycle, particularly in an opposed piston (OP) configuration. This study presents the virtual development of a single-cylinder 2-stroke OP engine with a total displacement of 0.95 L, designed to deliver 25 kW at 3000 rpm. Thanks to its compact size, high thermal efficiency, robustness, modularity, and low manufacturing cost, this engine is intended for use either as an industrial power unit or in combination with electric motors in hybrid vehicles. The overarching goal of this project is to demonstrate that internal combustion engines can offer a practical and cost-effective alternative to hydrogen fuel cells without significant penalties in terms of efficiency and pollutant emissions. The design of this novel engine started from scratch, and both 1D and 3D CFD simulations were employed, with particular focus on optimizing the cylinder’s geometry and developing an efficient low-pressure injection system. The numerical methodology was based on state-of-the-art commercial codes, in line with established engineering practices. The numerical results indicated that the optimized engine configuration slightly surpasses the target performance, achieving 29 kW at 3000 rpm, while maintaining near-zero NOx emissions (<20 ppm) and high brake thermal efficiency (~40%) over a wide power range. Additionally, the cost of this engine is projected to be lower than an equivalent 4-stroke engine, due to fewer components (e.g., no cylinder head, poppet valves, or camshafts) and a lighter construction. Full article

Memristors are state-of-the-art, nano-sized, two-terminal, passive electronic elements with very good switching and memory characteristics. Owing to their very low power usage and a good compatibility to the existing CMOS ultra-high-density integrated circuits and chips, they are potentially applicable in artificial and spiking neural networks, memory arrays, and many other devices and circuits for artificial intelligence. In this paper, a complete electronic realization of an analog circuit model of the modified neural net with memristor-based synapses and transfer function with memristors and MOS transistors in LTSPICE is offered. Each synaptic weight is realized by only one memristor, providing enormously reduced circuit complexity. The summing and scaling implementation is founded on op-amps and memristors. The logarithmic-sigmoidal activation function is based on a simple scheme with MOS transistors and memristors. The functioning of the suggested memristor-based neural network for pulse input signals is evaluated both analytically in MATLAB-SIMULINK and in the LTSPICE environment. The obtained results are compared one to another and are successfully verified. The realized memristor-based neural network is an important step towards the forthcoming design of complex memristor-based neural networks for artificial intelligence, for implementation in very high-density integrated circuits and chips. Full article

The proliferation of radar technology has given rise to a growing demand for advanced, high-performance transmitter front-ends operating in the 24 GHz frequency band. This paper presents a design analysis of a radio frequency (RF) transmitter (TX) front-end operated at a 24 GHz frequency and designed using 65 nm complementary metal-oxide-semiconductor (CMOS) technology for radar applications. The proposed TX front-end design includes the integration of an up-conversion mixer and power amplifier (PA). The up-conversion mixer is a Gilbert cell-based design that translates the 2.4 GHz intermediate frequency (IF) signal and 21.6 GHz local oscillator (LO) signal to the 24 GHz RF output signal. The mixer is designed with a novel technique that includes a duplex transconductance path (D_TP) for enhancing the mixer’s linearity. The D_TP of the mixer includes a primary transconductance path (P_TP) and a secondary transconductance path (S_TP). The P_TP incorporates a common source (CS) amplifier, while the S_TP incorporates an improved cross-quad transconductor (I_CQT). The integrated PA in the TX front-end is a class AB tunable two-stage PA that can be tuned with the help of varactors as a synchronous mode to increase the PA bandwidth or stagger mode to obtain a high gain. The PA is tuned to 24 GHz as a synchronous mode PA for the TX front-end operation. The proposed TX front-end showed an excellent output power of 11.7 dBm and dissipated 7.5 mW from a 1.2 V supply. In addition, the TX front-end achieved a power-added efficiency (PAE) of 47% and 1 dB compression point (OP₁dB) of 10.5 dBm. In this case, the output power is 10.5 dBm higher than the linear portion of the response. The methodologies presented herein have the potential to advance the state of the art in 24 GHz radar technology, fostering innovations in fields such as autonomous vehicles, industrial automation, and remote sensing. Full article

The paper deals with a brief overview of magnetoelastic sensors and magnetoelastic sensors used in general for sensing bending forces, either directly or sensing bent structures, and defines the current state of the art. Bulk magnetoelastic force sensors are usually manufactured from transformer sheets or amorphous alloys. In praxis, usually, a compressive force is sensed by bulk magnetoelastic sensors; however, in this paper, the sensor is used for the measurement of bending forces, one reason being that the effect of such forces is easily experimentally tested, whereas compressive forces acting on a single sheet make buckling prevention a challenge. The measurement of the material characteristics that served as inputs into a FEM simulation model of the sensor is presented and described. The used material was considered to be mechanically and magnetically isotropic and magnetically nonlinear, even though the real sheet showed anisotropic behavior to some degree. A sinusoidal magnetizing current waveform was used in the experimental part of this paper, which was created by a current source. The effects of various frequencies, amplitudes, and sensor geometries were tested. The experimental part of this paper studies the sensors’ RMS voltage changes to different loadings that bend the sheet out of its plane. The output voltage was the induced voltage in the secondary coil and was further analyzed to compute the linearity and sensitivity of the sensor at the specific current characteristic. It was found that for the given material, the most favorable operating conditions are obtained with higher frequency signals and higher excitation current amplitudes. The linearity of the sensor can be improved by placing the holes of the windings at different angles than 90° and by placing them further apart along the sheet’s length. The current source was created by a simple op-amp voltage-to-current source controlled by a signal generator, which created a stable waveform. It was found that transformer sheet bending sensors with the dimensions described in this paper are suitable for the measurement of small forces in the range of up to 2 N for the shorter sensors and approximately 0.2 N for the longer sensors. Full article

The airborne hyperspectral remote sensing systems (AHRSSs) acquire images with high spectral resolution, high spatial resolution, and high temporal dimension. While the AHRSS captures more detailed information from the terrain objects, the computational complexity of data processing is greatly increased. As an important application technology in the hyperspectral domain, anomaly detection (AD) processing must be real-time and high-precision in many cases, such as post-disaster rescue, military battlefield search, and natural disaster detection. In this paper, the real-time AD technology for the push-broom AHRSS is studied, the mathematical model is established, and a novel implementation framework is proposed. Firstly, the optimized kernel minimum noise fraction (OP-KMNF) transformation is employed to extract informative and discriminative features between the background and anomalies. Secondly, the Nyström method is introduced to reduce the computational complexity of OP-KMNF transformation by decomposing and extrapolating the sub-kernel matrix to estimate the eigenvector of the entire kernel matrix. Thirdly, the extracted features are transferred to hard disks for data storage. Then, taking the extracted features as input data, the background separation model-based CEM anomaly detector (BSM-CEMAD) is imported to detect anomalies. Finally, graphics processing unit (GPU) parallel computing is utilized in the Nyström-based OP-KMNF (NOP-KMNF) transformation and the BSM-CEMAD to improve the execution efficiency, and the real-time AD for the push-broom AHRSS could be realized. To test the feasibility of the implementation framework proposed in this paper, the experiment is carried out with the Airborne Multi-Modular Imaging Spectrometer (AMMIS) developed by the Shanghai Institute of Technical Physics as the data acquisition platform. The experimental results show that the proposed method outperforms many other state-of-the-art AD methods in anomalies detection and background suppression. Moreover, under the condition that the downlink data could retain most of the hyperspectral data information, the proposed method achieves real-time detection of pixel-level anomalies, with the initial delay not exceeding 1 s, the false alarm rate (FAR) less than 5%, and the true positive rate (TPR) close to 98%. Full article

The newest and upcoming geostationary passive imagers have thermal infrared channels comparable to those of more established instruments, but their spectral response functions still differ significantly. Therefore, retrievals developed for a certain type of radiometer cannot simply be applied to another imager. Here, a set of spectral band adjustment factors is determined for MSG/SEVIRI, Himawari-8/AHI, and MTG1/FCI from a training dataset based on MetOp/IASI hyperspectral observations. These correction functions allow to turn the observation of one sensor into an analogue observation of another sensor. This way, the same satellite retrieval—that has been usually developed for a specific instrument with a specific spectral response function—can be applied to produce long time series that go beyond one single satellite/satellite series or to cover the entire geostationary ring in a consistent way. It is shown that the mean uncorrected brightness temperature differences between corresponding channels of two imagers can be >1 K, in particular for the channels centered around 13.4 μm in the carbon dioxide absorption band and even when comparing different imager realizations of the same series, such as the four SEVIRI sensors aboard MSG1 to MSG4. The spectral band adjustment factors can remove the bias and even reduce the standard deviation in the brightness temperature difference by more than 80%, with the effect being dependent on the spectral channel and the complexity of the correction function. Further tests include the application of the spectral band adjustment factors in combination with (a) a volcanic ash cloud retrieval to Himawari-8/AHI observations of the Raikoke eruption 2019 and a comparison to an ICON-ART model simulation, and (b) an ice cloud retrieval to simulated MTG1/FCI test data with the outcome compared to the retrieval results using real MSG3/SEVIRI measurements for the same scene. Full article

Knowledge graphs are a popular research field in artificial intelligence, and store large amounts of real-world data. Since data are enriched over time, the knowledge graph is often incomplete. Therefore, knowledge graph completion is particularly important as it predicts missing links based on existing facts. Currently, the family of translation models delivers a better performance in knowledge graph completion. However, most of these models randomly generate negative triplets during the training process, resulting in the low quality of negative triplets. In addition, such models ignore the important characteristics of triplet-mapping properties during model learning. Therefore, we propose an optimization framework based on the translation models (Op-Trans). It enhances the knowledge-graph completion effect from both negative sampling and triplet-mapping properties. First, we propose a clustering cache to generate negative triplets, which generate negative triplets based on entity similarity. This sampling method can directly use the cache to track the negative triplets with large scores. In addition, we focus on the different contributions of the triplets to the optimization goal. We calculate the distinct weight for each triplet according to its mapping properties. In this way, the scoring function deals with each triplet depending on its own weight. The experimental results show that Op-Trans can help the state-of-the-art baselines to obtain a better performance in a link prediction task. Full article

Recent years have seen a significant increase in the scientific literature related to various methods for analyzing oxidative potential (OP) of atmospheric particulate matter (PM). The presence of several types of PM, differing chemical and physical properties, released by both anthropogenic and natural sources, leads to numerous health issues in living organisms and represents an attractive target for air quality monitoring. Therefore, several studies have focused on developing rapid and self-operative tests, employing different target molecules to assess OP of atmospheric aerosols as well as unique approaches to overcome some of the most common laboratory-related issues in this kind of analysis. This work provides an overview of online and automated systems, as well as a broad picture of the state-of-art of the various devices and methods developed on this topic over the last two decades. Moreover, representative studies on this subject will be discussed, analyzing the advantages and drawbacks of the developed automated techniques. Full article

Decomposition-based many-objective evolutionary algorithms (D-MaOEAs) are brilliant at keeping population diversity for predefined reference vectors or points. However, studies indicate that the performance of an D-MaOEA strongly depends on the similarity between the shape of the reference vectors (points) and that of the PF (a set of Pareto-optimal solutions symbolizing balance among objectives of many-objective optimization problems) of the many-objective problem (MaOP). Generally, MaOPs with expected PFs are not realistic. Consequently, the inevitable weak similarity results in many inactive subspaces, creating huge difficulties for maintaining diversity. To address these issues, we propose a two-state method to judge the decomposition status according to the number of inactive reference vectors. Then, two novel reference vector adjustment strategies, set as parts of the environmental selection approach, are tailored for the two states to delete inactive reference vectors and add new active reference vectors, respectively, in order to ensure that the reference vectors are as close as possible to the PF of the optimization problem. Based on the above strategies and an efficient convergence performance indicator, an active reference vector-based two-state dynamic decomposition-base MaOEA, referred to as ART-DMaOEA, is developed in this paper. Extensive experiments were conducted on ART-DMaOEA and five state-of-the-art MaOEAs on MaF1-MaF9 and WFG1-WFG9, and the comparative results show that ART-DMaOEA has the most competitive overall performance. Full article

The main challenge for constrained many-objective optimization problems (CMaOPs) is how to achieve a balance between feasible and infeasible solutions. Most of the existing constrained many-objective evolutionary algorithms (CMaOEAs) are feasibility-driven, neglecting the maintenance of population convergence and diversity when dealing with conflicting objectives and constraints. This might lead to the population being stuck at some locally optimal or locally feasible regions. To alleviate the above challenges, we proposed a dual-population-based NSGA-III, named DP-NSGA-III, where the two populations exchange information through the offspring. The main population based on the NSGA-III solves CMaOPs and the auxiliary populations with different environment selection ignore the constraints. In addition, we designed an $\epsilon$-constraint handling method in combination with NSGA-III, aiming to exploit the excellent infeasible solutions in the main population. The proposed DP-NSGA-III is compared with four state-of-the-art CMaOEAs on a series of benchmark problems. The experimental results show that the proposed evolutionary algorithm is highly competitive in solving CMaOPs. Full article

In the Software Development Life Cycle (SDLC), Development and Operations (DevOps) has been proven to deliver reliable, scalable software within a shorter time. Due to the explosion of Machine Learning (ML) applications, the term Machine Learning Operations (MLOps) has gained significant interest among ML practitioners. This paper explains the DevOps and MLOps processes relevant to the implementation of MLOps. The contribution of this paper towards the MLOps framework is threefold: First, we review the state of the art in MLOps by analyzing the related work in MLOps. Second, we present an overview of the leading DevOps principles relevant to MLOps. Third, we derive an MLOps framework from the MLOps theory and apply it to a time-series forecasting application in the hourly day-ahead electricity market. The paper concludes with how MLOps could be generalized and applied to two more use cases with minor changes. Full article

Dynamic Multi-Objective Optimization Problems (DMOPs) and Many-Objective Optimization Problems (MaOPs) are two classes of the optimization field that have potential applications in engineering. Modified Multi-Objective Evolutionary Algorithms hybrid approaches seem to be suitable to effectively deal with such problems. However, the standard Crow Search Algorithm has not been considered for either DMOPs or MaOPs to date. This paper proposes a Distributed Bi-behaviors Crow Search Algorithm (DB-CSA) with two different mechanisms, one corresponding to the search behavior and another to the exploitative behavior with a dynamic switch mechanism. The bi-behaviors CSA chasing profile is defined based on a large Gaussian-like Beta-1 function, which ensures diversity enhancement, while the narrow Gaussian Beta-2 function is used to improve the solution tuning and convergence behavior. Two variants of the proposed DB-CSA approach are developed: the first variant is used to solve a set of MaOPs with 2, 3, 5, 7, 8, 10,15 objectives, and the second aims to solve several types of DMOPs with different time-varying Pareto optimal sets and a Pareto optimal front. The second variant of DB-CSA algorithm (DB-CSA-II) is proposed to solve DMOPs, including a dynamic optimization process to effectively detect and react to the dynamic change. The Inverted General Distance, the Mean Inverted General Distance and the Hypervolume Difference are the main measurement metrics used to compare the DB-CSA approach to the state-of-the-art MOEAs. The Taguchi method has been used to manage the meta-parameters of the DB-CSA algorithm. All quantitative results are analyzed using the non-parametric Wilcoxon signed rank test with 0.05 significance level, which validated the efficiency of the proposed method for solving 44 test beds (21 DMOPs and 23 MaOPS). Full article