Mean Precision Research Articles

Paddy Yield Prediction in Tamilnadu Delta Region Using MLR-LSTM Model

ABSTRACT Crop yield forecasting has been well studied in recent decades and is significant in protecting food security. Crop growth is a complex phenomenon that depends on various factors. Machine learning and deep learning trends have emerged as important innovations in this field. We propose to utilize crop, weather, and soil data from agricultural datasets to evaluate yield prediction behavior. Paddy being a staple food crop in India is chosen for this research. In this paper, we propose hybrid architecture for paddy yield prediction, namely, MLR-LSTM, which combines Multiple Linear Regression and Long Short-Term Memory to utilize their complementary nature. The results are compared with traditional machine learning methods such as Support vector machine, Long short-term memory and Random forest method. Evaluation metrics such as Coefficient of Determination (R2), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Square Error (MSE), F1 score, Recall, and Precision are used to evaluate the hybrid method and traditional models. The results obtained from the proposed hybrid method indicates that the hybrid model delivers better R2, RMSE, MAE, MSE values of 0.93, 0.1549, 0.199, and 0.024 respectively.

Laparoscopic Video Analysis Using Temporal, Attention, and Multi-Feature Fusion Based-Approaches.

Adapting intelligent context-aware systems (CAS) to future operating rooms (OR) aims to improve situational awareness and provide surgical decision support systems to medical teams. CAS analyzes data streams from available devices during surgery and communicates real-time knowledge to clinicians. Indeed, recent advances in computer vision and machine learning, particularly deep learning, paved the way for extensive research to develop CAS. In this work, a deep learning approach for analyzing laparoscopic videos for surgical phase recognition, tool classification, and weakly-supervised tool localization in laparoscopic videos was proposed. The ResNet-50 convolutional neural network (CNN) architecture was adapted by adding attention modules and fusing features from multiple stages to generate better-focused, generalized, and well-representative features. Then, a multi-map convolutional layer followed by tool-wise and spatial pooling operations was utilized to perform tool localization and generate tool presence confidences. Finally, the long short-term memory (LSTM) network was employed to model temporal information and perform tool classification and phase recognition. The proposed approach was evaluated on the Cholec80 dataset. The experimental results (i.e., 88.5% and 89.0% mean precision and recall for phase recognition, respectively, 95.6% mean average precision for tool presence detection, and a 70.1% F1-score for tool localization) demonstrated the ability of the model to learn discriminative features for all tasks. The performances revealed the importance of integrating attention modules and multi-stage feature fusion for more robust and precise detection of surgical phases and tools.

Evaluating the Scalability of Matrix Factorization and Neighborhood Based Recommender Systems

Recommendation Systems are everywhere, from offline shopping malls to major e-commerce websites, all use recommendation systems to enhance customer experience and grow profit. With a growing customer base, the requirement to store their interest, behavior and respond accordingly requires plenty of scalability. Thus, it is very important for companies to select a scalable recommender system, which can provide the recommendations not just accurately but with low latency as well. This paper focuses on the comparison between the four methods KMeans, KNN, SVD, and SVD++ to find out the better algorithm in terms of scalability. We have analyzed the methods on different parameters i.e., Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Precision, Recall and Running Time (Scalability). Results are elaborated such that selection becomes quite easy depending upon the user requirements.

DFSP: A fast and automatic distance field-based stem-leaf segmentation pipeline for point cloud of maize shoot.

The 3D point cloud data are used to analyze plant morphological structure. Organ segmentation of a single plant can be directly used to determine the accuracy and reliability of organ-level phenotypic estimation in a point-cloud study. However, it is difficult to achieve a high-precision, automatic, and fast plant point cloud segmentation. Besides, a few methods can easily integrate the global structural features and local morphological features of point clouds relatively at a reduced cost. In this paper, a distance field-based segmentation pipeline (DFSP) which could code the global spatial structure and local connection of a plant was developed to realize rapid organ location and segmentation. The terminal point clouds of different plant organs were first extracted via DFSP during the stem-leaf segmentation, followed by the identification of the low-end point cloud of maize stem based on the local geometric features. The regional growth was then combined to obtain a stem point cloud. Finally, the instance segmentation of the leaf point cloud was realized using DFSP. The segmentation method was tested on 420 maize and compared with the manually obtained ground truth. Notably, DFSP had an average processing time of 1.52 s for about 15,000 points of maize plant data. The mean precision, recall, and micro F1 score of the DFSP segmentation algorithm were 0.905, 0.899, and 0.902, respectively. These findings suggest that DFSP can accurately, rapidly, and automatically achieve maize stem-leaf segmentation tasks and could be effective in maize phenotype research. The source code can be found at https://github.com/syau-miao/DFSP.git.

METHOD DEVELOPMENT AND VALIDATION FOR QUANTIFICATION OF OZENOXACIN IN PHARMACEUTICAL PRODUCTS BY MICROBIOLOGICAL ASSAY

Ozenoxacin is a novel non-fluorinated topical antibiotic belonging to the quinolone family. HPLC techniques are normally used for the quantification of the efficacy of ozenoxacin, however, the HPLC technique is unable to infer bioactivity against microorganisms. The goal of this article was to augment and authenticate an accurate, sensitive, and yet simple, commercial two-stage cylinder plate (2+2) assay for determining the efficacy and bioactivity of ozenoxacin in pharmaceutical products, which has not been published in any of the scientific journals. This study is based on the development of a bioassay technique for estimating ozenoxacin in pharmaceutical products. In this investigation, the microbiological assay was performed with 08 bacterial strains and 02 fungal strains, from which Salmonella enterica serotype Abony MTCC-3858 was selected as the most suitable organism against ozenoxacin based on sharp and clear zone of inhibition. The advanced mathematical approach to linearity, precision and accuracy has been successfully verified and validated. The mean potency recovery value was estimated to be 100.83 % for ozenoxacin cream. In this study, validated methods were used to assess the linearity (r2 =0.9859); the intermediate precision RSD between days was 1.01 %; The mean precision RSD between analysts was 1.04 % and accuracy was 100.67 %, RSD = 0.65 %. Bioassay was optimized by investigating several factors such as robustness by buffer diluent used for standard and sample preparation (phosphate buffer pH 8.0; RSD=0.37 %), inoculum concentration (0.3 %; RSD=0.44 %) and incubation temperature (30 o C; RSD=0.80 %). The result shows that the efficacy of ozenoxacin in pharmaceutical products can be estimated by bioassay technique for quality control purpose.

A water quality assessment method based on an improved grey relational analysis and particle swarm optimization multi-classification support vector machine.

Most of the water quality indicators that affect the results of river water quality assessment are gray and localized, thus the correlation between water quality indicators can be calculated using gray correlation analysis (GRA).However, GRA takes equal weighting for water quality indicators and does not take into account the weighting of the indicators. Therefore, this paper proposes a river water quality assessment method based on improved grey correlation analysis (ACGRA) andparticle swarm optimization multi-classification support vector machine (PSO-MSVM) for assessing river water environment quality. Firstly, the combination weights of water quality indicators were calculated using Analytic Hierarchy Process (AHP)AHP and Criteria Importance Though Intercrieria Correlation (CRITIC)CRITIC, and then the correlation between water quality indicators was calculated for feature selection. Secondly, the PSO-MSVM model was established using the water quality indicators obtained by ACGRA as input parameters for water environment quality assessment. The river water environment assessment methods of ACGRA and PSO-MSVM were applied to the evaluation of water environment quality in different watersheds in the country. Accuracy, precision, recall and root mean square errorRMSE were also introduced as model evaluation criteria. The results show that the river water environment assessment methods based on ACGRA and PSO-MSVM can evaluate the water environment quality more accurately.

Mapping Coastal vulnerability using machine learning algorithms: A case study at North coastline of Sebou estuary, Morocco

The coastline of Morocco has severe fragility presented in coastal vulnerability to natural and anthropogenic impacts. Our work aims first to analyze the coastal vulnerability using the CVI index can highlight the potential vulnerabilities associated with different land uses and identify hot spots (accretion/erosion) in the northern area of the estuary of Sebou’, Morocco. The northern area of Sebou estuary experiences high vulnerability since it has a low topography, which in turn leads to different coastal issues (erosion/accretion).The research employs a three-step process to achieve its goal. First, shoreline location was extracted from several satellite images (Landsat TM+ and ETM+) during the last 19 years. Moreover, the rates of shoreline change were extracted by integrating ArcGIS and DSAS. Finally, various machine learning algorithms consisting of support vector machine (SVM), random forest (RF), decision tree (DT), logistic regression (LR), and artificial neural networks were used to map coastal vulnerability (ANN). A total of 4013 points were used to differentiate between high and low vulnerability classes. Based on data derived from DEM, DSAS, and other sources, the computations took into account 8 predictors of coastal vulnerability (elevation, slope, geomorphology, natural habitat, maximum wave height, sea level rise, tidal range, and shoreline change).The obtained data were divided into 70% for model training and 30% for testing. For the evaluation of the proposed models, several measures were used, namely: precision, recall, F1 score, kappa index, and root mean square error (RMSE). The result showed that RF (precision = 99%, Kappa 0.98), DT (precision = 99%, Kappa index = 0.99), and SVM (precision = 88%, Kappa index = 0.83) outperformed the other methods. The prediction was overlaid on (≈76 km) of the coastline. It was observed that about 20.07% (≈15.25 km) of the coastline is moderately vulnerable, while 10.35% (≈7.86 km) is highly vulnerable and 69.57% (≈52.87 km) has a low vulnerability.The suggested method can actively contribute to the sustainable management of the coastal area and is a useful tool for forecasting coastal vulnerability. At the same time, it aids in planning, designing, and facilitating the implementation of adaption strategies by decision-makers.

Automated CT angiography collateral scoring in anterior large vessel occlusion stroke: A multireader study.

Computed tomography (CT) angiography collateral score (CTA-CS) is an important clinical outcome predictor following mechanical thrombectomy for ischemic stroke with large vessel occlusion (LVO). The present multireader study aimed to evaluate the performance of e-CTA software for automated assistance in CTA-CS scoring. Brain CTA images of 56 patients with anterior LVO were retrospectively processed. Twelve readers of various clinical training, including junior neuroradiologists, senior neuroradiologists, and neurologists graded collateral flow using visual CTA-CS scale in two sessions separated by a washout period. Reference standard was the consensus of three expert readers. Duration of reading time, inter-rater reliability, and statistical comparison of readers' performance metrics were analyzed between the e-CTA assisted and unassisted sessions. e-CTA assistance resulted in significant increase in mean accuracy (58.6% to 67.5%, p = 0.003), mean F1 score (0.574 to 0.676, p = 0.002), mean precision (58.8% to 68%, p = 0.007), and mean recall (58.7% to 69.9%, p = 0.002), especially with slight filling deficit (CTA-CS 2 and 3). Mean reading time was reduced across all readers (103.4 to 59.7 s, p = 0.001), and inter-rater agreement in CTA-CS assessment was increased (Krippendorff's alpha 0.366 to 0.676). Optimized occlusion laterality detection was also noted with mean accuracy (92.9% to 96.8%, p = 0.009). Automated assistance for CTA-CS using e-CTA software provided helpful decision support for readers in terms of improving scoring accuracy and reading efficiency for physicians with a range of experience and training backgrounds and leading to significant improvements in inter-rater agreement.

A Fast Ground Segmentation Method of LiDAR Point Cloud From Coarse-to-Fine

Light detection and ranging (LiDAR) has become an indispensable sensor for autonomous vehicles due to its unique properties. Segmentation of raw point cloud is a fundamental step in the perception processing pipeline, which is critical to the safe driving of autonomous vehicles. However, some existing methods are prone to over-segmentation, insufficient segmentation, or poor real-time performance in the process of point cloud segmentation. Therefore, we propose a coarse-to-fine segmentation strategy to balance segmentation accuracy and speed. The ground segmentation includes applying robust locally weighted regression (RLWR) to the nonuniformly divided bins of polar grids in concentric model to achieve coarse ground segmentation. The processing results are mapped to a range image, and the improved range image convolution method is used to refine the ground segmentation. We evaluate the proposed method on two public datasets, and the precision, recall, and harmonic mean of the ground segmentation are all above 93%, which is sufficient to demonstrate the robustness of our method. Using the segmentation strategy from coarse to fine, the average processing time per frame of the SemanticKITTI dataset and the Nagoya dataset are 39.49 and 42.63 ms, respectively, with an average of 41.06 ms, which can meet the real-time requirements of autonomous driving.

Prediction of the severity of Dothistroma needle blight in radiata pine using plant based traits and narrow band indices derived from UAV hyperspectral imagery

Dothistroma needle blight, caused by the fungi Dothistroma septosporum and D. pini is globally one of the most damaging diseases of pine species. Infection from the pathogen, which is initiated in the lower crown, causes needle chlorosis, loss of cellular integrity, necrosis and premature needle loss with this developmental sequence impacting key plant functional traits such as chlorophyll content and leaf area index. This study uses data collected from four year old field grown radiata pine (Pinus radiata D. Don) exhibiting a wide range in disease severity (0 – 58%). Hyperspectral data collected from a UAV was used to determine how plant functional traits inverted from a 3D radiative transfer model (PRO4SAIL), can be used by random forests to improve predictions of disease severity, relative to a model with only narrowband hyperspectral indices. Model precision was assessed on a test dataset that was not used for model fitting and this cross validation process was repeated 50 times to ensure reported results were robust, with the mean precision reported from these repeats.The model of disease severity using narrow band hyperspectral indices was moderately precise (R2 = 0.52). Addition of the inverted plant functional traits markedly improved model precision and this final model strongly predicted disease severity with R2 of 0.85. Analyses of variable importance in the final model showed significant interchange in variable importance as the disease progressed. Early stages of the disease were most readily distinguished from asymptomatic trees (severity = 0%) using variables that predominantly describe changes in photosynthesis (Photochemical Reflectance Index), chlorophyll degradation (Normalized Phaeophytinization Index) and chlorophyll content. In contrast, the more severe impacts of the disease were most well characterised by traits associated with an extreme reduction in transpiration (carotenoid content) and loss of foliage.

MUESTREO SECUENCIAL PARA EL MONITOREO DE RENDIMIENTOS EN GRANJAS DEL GUSANO REY (ZOPHOBAS SPP. [COLEOPTERA: TENEBRIONIDAE]) EN LA AMAZONIA

<p class="rfpMnscText"><strong>Background</strong>: Sampling for monitoring commercial production of kingworm (<em>Zophobas</em> spp.) requires knowledge about optimal sample sizes and methods for projecting yields. <strong>Hypothesis and objective</strong>: A sequential enumerative sampling (SES) based on larval weight records should present significant levels of productivity prediction under known margins of error and precision. To verify this, it was proposed to evaluate the performance of an SES method in two kingworm’s farms in Tarapoto, Peruvian Amazon. <strong>Methodology</strong>: Means (m) and respective variances (s<sup>2</sup>) of larvae number and weight since six sample units (SU) of 500 ml from 35 production units (PU) of 48 l were obtained. Log-transformed data were fitted to Taylor’s Power Law TPL (log s<sup>2</sup> =log a +b. log m). A Morisita’s index transformation was applied with original larvae number data to obtain an independent measure of intra-sample spatial arrangement. TPL elements constituted the Green's function optimal sample size for three margins of error E. Sequential sampling simulations were carried out, whose means predicted yields comparing two methods of weight relationship between SU and PU (w/W, linear function), with their respective census. <strong>Results</strong>: Number of larvae (s<sup>2</sup> =1.06 m<sup>1.61</sup>) and weights (s<sup>2</sup> =0.79 m<sup>1.53</sup>) showed highly significant adjustments in TPL, with aggregation coefficients corresponding crowded spatial arrangement. Larval density and I<sub>p</sub> index showed close correlation. Although simulations provided a lower hit frequency than expected, the mean precision increased highly significantly while increasing the error margin, with detection levels of 0.25, 2.23 and 12.57 g to E 50, 30 and 20 %, respectively. <strong>Implications</strong>: Applying SES, kingworm’s breeders should standardize substrate volume contained in each UP and adjust their w/W conversion factor to avoid plus sampling efforts. <strong>Conclusion</strong>: The SES suits the needs of productivity monitoring, where using of w/W ratio and 50 % margins of error are associated with greater effectiveness and precision with a sample size of n<sub>i</sub><3. </p>

Mesh-Based DGCNN: Semantic Segmentation of Textured 3-D Urban Scenes

Textured 3D mesh is one of the final user products in photogrammetry and remote sensing. However, research on the semantic segmentation of complex urban scenes represented by textured 3D meshes is in its infancy. We present a mesh-based dynamic graph CNN (DGCNN) for the semantic segmentation of textured 3D meshes. To represent each mesh facet, composite input feature vectors are constructed by concatenating the face-inherent features, i.e., XYZ coordinates of the center of gravity (CoG), texture values, and normal vectors. A texture fusion module is embedded into the proposed mesh-based DGCNN to generate high-level semantic features of the high-resolution texture information, which is useful for semantic segmentation. We achieve competitive accuracies when the proposed method is applied to the SUM mesh datasets. The overall accuracy (OA), Kappa coefficient (Kap), mean precision (mP), mean recall (mR), mean F1 score (mF1), and mean intersection over union (mIoU) are 93.3%, 88.7%, 79.6%, 83.0%, 80.7%, and 69.6%, respectively. In particular, the OA, mean class accuracy (mAcc), mIoU, and mF1 increase by 0.3%, 12.4%, 3.4%, and 6.9%, respectively, compared to the state-of-the-art method.

Internal Wave Signature Extraction From SAR and Optical Satellite Imagery Based on Deep Learning

Internal waves (IWs) are a common characteristic of oceans and serve a crucial role in transmitting energies between large-scale tides and small-scale mixing. This study developed a deep-learning-based method for extracting IW signatures on multiple satellite imagery from synthetic aperture radar (SAR) and optical sensors in sun-synchronous and geostationary orbits with varying spatial resolution. We collected 1115 satellite images, including 116 ENVISAT ASAR (Advanced SAR), 839 MODIS (MODerate-resolution Imaging Spectroradiometer), and 160 Himawari-8 AHI (Advanced Himawari Imager) images with clear IW signatures in the South China Sea (SCS), Sulu Sea, and Celebes Sea for model training. Considering the distinct IW characteristics under different imaging mechanisms, the specially tailored IW Extraction network (IWE-Net) leverages three modifications to improve the accuracy and robustness: online data augmentation, squeeze and excitation blocks, and Matthews correlation coefficient loss. The overall mean Precision, Recall, and F1-score of the IWE-Net model are 85.75%, 85.67%, and 85.71%, demonstrating the model is accurate for IW signature extraction.We also proved the transferability of our method to sea areas worldwide, long-term periods, and Sentinel satellite sensors completely independent of the model training. Globally, the number of IW images and extracted pixels show an obvious tidal-related double-peak distribution. Furthermore, we processed 15461 MODIS images in the northeastern SCS to present a holistic IW distribution map over the past 22 years. An unreported IW silent zone caused by drastic topography changes has been discovered, indicating the great potential of deep learning in information retrieval from remote sensing imagery.

MESAC: Learning to Remove Mismatches via Maximizing the Expected Score of Sample Consensuses

Most learning-based methods require labelling the training data, which is time-consuming and gives rise to wrong labels. To address the labelling issues thoroughly, we propose an unsupervised learning framework to remove mismatches by maximizing the expected score of sample consensuses (MESAC). The proposed MESAC can train various permutation invariant networks (PINs) based on training data with no labels, and has three distinct merits: 1) the framework can train various PINs in an unsupervised mode such that these are immune to wrong labels; 2) the gradients of the expected score are explicitly calculated by a revised score-function estimator, which can avoid gradient explosion; 3) the distribution of matching probabilities is learned from the PIN and precisely modelled by a categorical distribution, which can decrease the sampling times and improve the computational efficiency accordingly. Experiments of testing datasets disclose that mean recall is increased by at most 77% when pure PINs are embedded in MESAC, and mean precision is also improved by 16%. Applications in pose recovery indicate that the success rates of MESAC-integrated PINs outperform the compared methods when training with neither matching labels nor ground truth epipolar geometry (EG) constraints, showing the great potential of MESAC in mismatch removal.

Anchor-Free Feature Aggregation Network for Instrument Detection in Endoscopic Surgery

Endoscopic endonasal approach has been widely used for removing various sellae tumors including pituitary adenomas, meningiomas, etc. While, performing these surgeries in such a narrow space with different instruments remains a challenge for surgeons, due to the limited field of view, varying illumination, and occlusion of instruments during the operation. Thus, a proper surgical instrument detection method that can provide classification and location information of the operated surgical instrument is critical for surgeons to understand the surgical scenarios and enhance the safety of the clinical operation. To this end, we propose an anchor-free feature aggregation network (AFA-Net) to improve the detection precision of surgical instruments from the endoscopic operation view field. The proposed method utilizes the improved feature pyramid network (FPN) with the depthwise separable convolution and a weighted feature aggregation module to enhance the feature information of the operated surgical instruments. Based on the anchor-free method, a weighted heatmap aggregation module is used to detect surgical instruments. Experimental studies on a public dataset Cholec80 and an intraoperative dataset from a local hospital are conducted, and the detection performance is assessed by the mean precision (AP) and average recall (AR). From both datasets and comparisons, the proposed method achieves 74.1% AP, 67.0% AR and 73.6% AP, 66.7% AR, respectively, which show significant advantages over five mainstream methods in terms of detection performance.

Kinematic Calibration of a 6-DoF Parallel Manipulator With Random and Less Measurements

Parallel manipulator (PM) embraces more kinematic errors than serial mechanism. The substantial errors in PM usually demand more measurements, which leads to low efficiency and affects the accuracy of identification due to the ill-conditioning problem. Having realized this common problem in the kinematic calibration of PMs, we carried out a finite and instantaneous screw (FIS)-based kinematic calibration on a 6- <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</u> UHU PM before application to fracture reduction. <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</u> , U, and H denote actuated revolute joint, universal joint, and helical joint, respectively. The error model is built by the error mapping of serial chains in the FIS framework. Instead of devoting to the measurement planning for more precise identification, the identifiability of kinematic errors is thoroughly analyzed. Redundant errors are categorized as the ones within limb and among limbs. A minimal identification model is thus defined and an error conversion method is proposed for compensation. Calibration experiments show that the FIS method has an improved accuracy of the 6- <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</u> UHU PM at least one magnitude higher than the precision demand, i.e., mean position and orientation errors 0.0574 mm and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$5.1197\,\,{\times }\,\, 10^{-4}$ </tex-math></inline-formula> rad after calibration. Compared with the conventional closed-loop vector (CLV) equation method, the accuracy improvement of FIS method is higher. In addition, comparisons with the commonly used configuration selection methods indicate that our method allows random and a smaller number of measurements and can reach satisfactory accuracy.

Combined Effect of Concept Drift and Class Imbalance on Model Performance During Stream Classification

Every application in a smart city environment like the smart grid, health monitoring, security, and surveillance generates non-stationary data streams. Due to such nature, the statistical properties of data changes over time, leading to class imbalance and concept drift issues. Both these issues cause model performance degradation. Most of the current work has been focused on developing an ensemble strategy by training a new classifier on the latest data to resolve the issue. These techniques suffer while training the new classifier if the data is imbalanced. Also, the class imbalance ratio may change greatly from one input stream to another, making the problem more complex. The existing solutions proposed for addressing the combined issue of class imbalance and concept drift are lacking in understating of correlation of one problem with the other. This work studies the association between concept drift and class imbalance ratio and then demonstrates how changes in class imbalance ratio along with concept drift affect the classifier’s performance. We analyzed the effect of both the issues on minority and majority classes individually. To do this, we conducted experiments on benchmark datasets using state-of-the-art classifiers especially designed for data stream classification. Precision, recall, F1 score, and geometric mean were used to measure the performance. Our findings show that when both class imbalance and concept drift problems occur together the performance can decrease up to 15%. Our results also show that the increase in the imbalance ratio can cause a 10% to 15% decrease in the precision scores of both minority and majority classes. The study findings may help in designing intelligent and adaptive solutions that can cope with the challenges of non-stationary data streams like concept drift and class imbalance.

MSGraph: Modeling multi-scale K-line sequences with graph attention network for profitable indices recommendation

&lt;abstract&gt;&lt;p&gt;Indices recommendation is a long-standing topic in stock market investment. Predicting the future trends of indices and ranking them based on the prediction results is the main scheme for indices recommendation. How to improve the forecasting performance is the central issue of this study. Inspired by the widely used trend-following investing strategy in financial investment, the indices' future trends are related to not only the nearby transaction data but also the long-term historical data. This article proposes the MSGraph, which tries to improve the index ranking performance by modeling the correlations of short and long-term historical embeddings with the graph attention network. The original minute-level transaction data is first synthesized into a series of K-line sequences with varying time scales. Each K-line sequence is input into a long short-term memory network (LSTM) to get the sequence embedding. Then, the embeddings for all indices with the same scale are fed into a graph convolutional network to achieve index aggregation. All the aggregated embeddings for the same index are input into a graph attention network to fuse the scale interactions. Finally, a fully connected network produces the index return ratio for the next day, and the recommended indices are obtained through ranking. In total, 60 indices in the Chinese stock market are selected as experimental data. The mean reciprocal rank, precision, accuracy and investment return ratio are used as evaluation metrics. The comparison results show that our method achieves state-of-the-art results in all evaluation metrics, and the ablation study also demonstrates that the combination of multiple scale K-lines facilitates the indices recommendation.&lt;/p&gt;&lt;/abstract&gt;

AFU-Net: A Novel U-Net Network for Rice Leaf Disease Segmentation

Highlights The attention mechanism enhances the ability of the model to learn specific semantic information in encoder. The redesigned residual structure deepens the network while reducing the number of parameters. The feature extraction module and feature fusion module obtain richer boundary feature information and effectively integrate output results from different levels. The mIoU, mPA, and Precision values of AFU-Net in the self-built dataset are 87.25%, 92.23%, and 99.67%, respectively. Abstract. Rice diseases adversely affect rice growth and yield. Precise spot segmentation helps to assess the severity of the disease so that appropriate control measures can be taken. In this article, we propose a segmentation method called AFU-Net for rice leaf diseases, and its performance is verified through experiments. Based on the traditional UNet, this method incorporates an attention mechanism, a residual module and a feature fusion module (FFM). The attention mechanism is embedded in skip connections, which enhances the learning of particular semantic features in the encoder layer. In addition, the residual module is integrated into the decoder layer, which deepens the network and enables it to extract richer semantic information. The proposed FFM structure effectively enhances the learning of boundary information and local detail features. The experimental results show that the mean intersection over union (mIoU), mean pixel accuracy (mPA) and Precision of the proposed model on the self-built rice leaf disease segmentation dataset are 87.25%, 92.23%, and 99.67%, respectively. All three evaluation indexes were improved over the control group, while the proposed model had the lowest number of parameters and displayed a good segmentation effect for smaller disease points and disease parts with less obvious characteristics. Keywords: Attention mechanism, Feature fusion module, Residual module, Rice leaves, UNet model.

Comprehensive Analysis of Nature-Inspired Algorithms for Parkinson’s Disease Diagnosis

Parkinson’s disease (PD) is a prominent neurodegenerative disease that damages the neurons of the substantia nigra, causing irreversible impairments leading to involuntary movements. As this disease disrupts patients’ daily activities in a mature stage, early detection of the disease is crucial. Several methods based on nature-inspired (NI) algorithms have been proposed for PD detection and patient management. As there are several NI algorithms for feature selection, a mapping with an individual machine learning (ML) classifier is necessary to obtain optimal performance of the detection pipeline. To fill this gap, in this work, 13 NI algorithms and 11 ML classifiers were selected, and critical comparisons were performed regarding their combined performance in detecting PD. Each NI algorithm was employed to select an optimal feature set which was then classified by the 11 ML classifiers keeping the same parameters. This generated 143 NI-ML pairs, which were carefully compared to find the best-performing pairs considering several assessment criteria such as accuracy, cross-validation mean score, precision, recall and F1-score. The results of the extensive comparative analysis allowed the ranking of the algorithms in the 50th, 75th and 95th percentile to identify the best-performing pairs. The analyses revealed that 12 NI-ML models obtained a testing accuracy of over 91%, which is above the 95th percentile value. The Flower Pollination Algorithm and Extreme Gradient Boost Algorithm pair obtained the highest testing accuracy of 93%. This study revealed the remarkable performance of the boosting algorithms promoting explainable machine learning in PD detection.