Using deep learning techniques in image analysis is a dynamically emerging field. This study aims to use convolutional neural network (CNN), approach, automatically classify esophageal cancer (EC) and distinguish it from premalignant lesions.A total of 1,272 white-light images were adopted 748 subjects, including normal cases, lesions, cancerous lesions; 1,017 used train the CNN, another 255 examined evaluate CNN architecture. Our proposed structure consists two subnetworks (O-stream...

The ability to decompose complex multi-object scenes into meaningful abstractions like objects is fundamental achieve higher-level cognition. Previous approaches for unsupervised object-oriented scene representation learning are either based on spatial-attention or scene-mixture and limited in scalability which a main obstacle towards modeling real-world scenes. In this paper, we propose generative latent variable model, called SPACE, that provides unified probabilistic framework combines...

Abstract Human Activity Recognition (HAR) has had a diverse range of applications in various fields such as health, security and smart homes. Among different approaches HAR, WiFi-based solutions are getting popular since it solves the problem deployment cost, privacy concerns restriction applicable environment. In this paper, we propose human activity recognition system that can identify activities via channel state information from WiFi devices. A special deep learning framework, Long...

Low temperature and soil salinization during cotton sowing seedling adversely affect productivity. Exogenous melatonin (MT) can alleviate the damage caused to plants under non-biological stress; thus, applying MT is a means improve growth condition of crops stress. However, achieving this goal requires thorough understanding physiological regulatory mechanisms on seedlings low salinity This study could bring new knowledge physio-biochemical that tolerance combined effects salt stress using...

Nowadays, Unmanned Aerial Vehicles (UAVs) have received growing popularity in the Internet-of-Things (IoT) which often deploys many sensors a relatively wide region. Since battery capacity is limited, cannot transmit over long distance. It necessary for designing efficient sensor data collection mechanisms to prolong lifetime of IoT and enhance efficiency. In this paper, we consider UAV-enabled scenario, where multiple heterogeneous UAVs with different energy constraints are employed collect...

Long−term excessive nitrogen fertilizer input has resulted in several environmental problems, including an increase N2O emissions and the aggravation of nitrate leaching; monitoring is crucial for maize with high yield. This study aimed to optimize amount applied by Climate−Smart Agriculture (CSA) so as continuously improve agricultural productivity reduce or eliminate much possible. Field experiments a completely randomized design were conducted examine effects six treatments (N application...

In the North China Plain, drip irrigation is gradually used in winter wheat production, and improper management of water fertilizer aggravates risk crop instability groundwater pollution. A lysimeter experiment with three levels fertilization timing (T1 = beginning; T2 middle; T3 end cycle) two rates (W1 30 mm W2 20 mm) was carried out to investigate effects rate on soil NO3−-N distribution, development, yield, water–nitrogen usage efficiency wheat. The results indicated that, under...

Highlights The proposed data enhancement method can be used for small-scale sets with rich sample image features. accuracy of the new model reaches 98.5%, which is better than traditional CNN method. Abstract : GoogLeNet offers far performance in identifying apple disease compared to methods. However, complexity relatively high. For small volumes data, does not achieve same as it large-scale data. We propose a identification using GoogLeNet’s inception module. adopts variety methods optimize...

When doctors assess the liver reserve function of patients with cancer before hepatectomy, if assessment is not appropriate, are prone to postoperative failure(POLF). Child-Pugh most commonly used method for clinical evaluation, but this only relies on biochemical indicators, which may lead poor prediction. This paper aims evaluate imaging omics information using an effective machine learning scheme. The performance models including support vector (SVM), COP-Kmeans, MLP, SGD-S3VM, QN-S3VM...

In this study, to solve the low accuracy of multi-space-based sensor joint tracking in presence unknown noise characteristics, an adaptive multi-sensor algorithm (AMSJTA) is proposed. First, coordinate transformation from target object optical sensors considered, and observation vector-based measurement model established. Then, characteristics are assumed be white Gaussian noise, covariance matrix set as a constant. On premise, traditional iterative extended Kalman filter applied problem....

Partial discharge (PD) surveillance constitutes a pivotal methodology for diagnosing insulation failures in electrical equipment. Enhancing comprehensively the precision of identifying PD anomalies Gas Insulated Switchgear (GIS) is paramount significance ensuring steady functioning power grids. This study introduces novel framework that integrates Phase-Resolved Graph Segmentation (PRPD-Grabcut) with tailored MobileNets-based Convolutional Neural Network (MCNN) to classify GIS-related...

This paper studies the relay selection in wireless cooperative communication systems. Since high mobility of Unmanned Aerial Vehicles (UAVs) will lead to change link state, a algorithm based on position prediction is proposed this better study problem. First, Unscented Kalman Filter (UKF) used predict real-time location information each UAV, and information, it determined whether current UAV node needs be replaced. The predicted by UKF algorithm, which basis for determining Then, outage...

Image registration is a research hotspot in computer vision. Depending on the overlapping region between adjacent images, this technology can complete task of matching and superposition multiple images. In reality, getting highly accurate performance complex motion states meeting realtime requirement for high-resolution images simultaneously still challenge. It observed that occupies only 30% to 50% Meanwhile, points are centralized region. Thus, detecting contribute reducing time improving...

Excessive nitrogen application would deteriorate soil structure and increase greenhouse gas emission. We set up six treatments, i.e., N0, N120, N180, N240, N300and N360(nitrogen rates of 0, 120, 180, 240, 300 360 kg·hm-2, all straws returned into the field in situ) fertilizer experimental site to investigate effects different on N2O emission, water-filled porosity (WFPS), temperature, nitrate ammonium contents, composition stability water stable aggregates winter wheat filed 2018-2020. The...

Spaceborne remote sensing data is one of the crucial sources in accurate measurement systems, which provides security for users to perform tasks practical applications. When BeiDou short-message communication (BDSMC) applied transmission spaceborne data, success rate and timeliness directly determine effectiveness data. However, BDSMC has several problems: a small amount transferred each time, process unreliable. To address these problems, this paper designs long message fragmentation...

With the rapid development of China's economy, cities are becoming more and prosperous, many lit up at night. However, when urban lighting exceeds a certain limit affects life residents, it becomes light pollution. In order to monitor prevent pollution, we first defined three kinds pollution accurately, constructed evaluation model for Shanghai based on ground-based satellite remote sensing technology sampling data, using advantages support vector machine robustness, can effectively avoid...

Leaf nitrogen content (LNC) is an important factor reflecting the growth quality of plants. We estimated apple leaves using hyperspectral wavelength analysis differential spectrum, spectrum transformation, and vegetation index with different derivative gaps. then used characteristic wavelengths extracted via correlation coefficient method as input vectors to gradient boosting decision tree (GBDT) model for performed cross-validation optimize inversion parameters. analyzed results variables...