Neural Network Learning Algorithm Research Articles

Quantized non-volatile nanomagnetic domain wall synapse based autoencoder for efficient unsupervised network anomaly detection

Abstract Anomaly detection in real-time using autoencoders implemented on edge devices is exceedingly challenging due to limited hardware, energy, and computational resources. We show that these limitations can be addressed by designing an autoencoder with low-resolution non-volatile memory-based synapses and employing an effective quantized neural network learning algorithm. We further propose nanoscale ferromagnetic racetracks with engineered notches hosting magnetic domain walls (DW) as exemplary non-volatile memory based autoencoder synapses, where limited state (5-state) synaptic weights are manipulated by spin orbit torque (SOT) current pulses to write different magnetoresistance states. The performance of anomaly detection of the proposed autoencoder model is evaluated on the NSL-KDD dataset. Limited resolution and DW device stochasticity aware training of the autoencoder is performed, which yields comparable anomaly detection performance to the autoencoder having floating-point precision weights. While the limited number of quantized states and the inherent stochastic nature of DW synaptic weights in nanoscale devices are typically known to negatively impact the performance, our hardware-aware training algorithm is shown to leverage these imperfect device characteristics to generate an improvement in anomaly detection accuracy (90.98%) compared to accuracy obtained with floating-point synaptic weights that are extremely memory intensive. Furthermore, our DW-based approach demonstrates a remarkable reduction of at least three orders of magnitude in weight updates during training compared to the floating-point approach, implying significant reduction in operation energy for our method. This work could stimulate the development of extremely energy efficient non-volatile multi-state synapse-based processors that can perform real-time training and inference on the edge with unsupervised data.

Syngas composition analysis for waste to methanol production: Techno-economic assessment using machine learning and Aspen plus

The yield of methanol is heavily influenced by the composition of the syngas which varies from waste type and gasification conditions. This study uniquely explores machine learning (ML) and artificial neural network (ANN) algorithms with shapley additive values (SHAP) analysis to predict syngas compositions and interpret the complexities of waste gasification. Aspen Plus is then utilized to investigate the impact of syngas compositions on the economics of methanol production. Among the different ML models, gradient boosting regression achieved remarkable accuracy in predicting carbon monoxide and carbon dioxide, while ANN excelled in predicting nitrogen and hydrogen based on the coefficient of determination (>0.95). The feature importance analysis demonstrated the gasification agent type as the most important feature. Steam gasification of waste sawdust, straw, and wood chips showed the highest methanol yield. Economic analysis indicated the lowest levelized cost of methanol (LCMeOH) to be 1195.79 USD/MT which further decreased to 933.97 USD/MT upon hydrogen addition. Uncertainty analysis using Monte Carlo method showed the possible minimum LCMeOH to be 254.05 USD/MT. This study provides a comprehensive picture of the economic viability of waste-to-methanol production and provides an important reference for stakeholders on a global scale to quickly assess the feasibility of waste-to-energy projects.

Intelligent Lighting Control Systems for Energy Savings in Hospital Buildings Using Artificial Neural Networks

Lighting control systems are essential in modern building automation and smart homes, efficiently managing illumination to enhance energy conservation and user comfort. This project tackles energy consumption challenges in hospital buildings by introducing Intelligent Lighting Control Systems (ILCS) that take natural light and occupancy into account, driven by Artificial Neural Networks (ANN) and diverse machine learning algorithms. In our study, we collected sensor data, processed it, and designed a lighting control system employing a feedforward neural network and various machine learning algorithms. Surprisingly, our research found that a linear regression algorithm surpassed the ANN-based system in this context. We implemented a prototype, tested it on hardware, and obtained the expected results. This research marks progress towards optimizing energy use in hospital buildings and contributing to sustainability endeavors. By combining ILCS and machine learning, it offers a promising approach for more efficient and eco-friendly lighting systems

Boosting Clear Cell Renal Carcinoma-Specific Drug Discovery Using a Deep Learning Algorithm and Single-Cell Analysis.

Clear cell renal carcinoma (ccRCC), the most common subtype of renal cell carcinoma, has the high heterogeneity of a highly complex tumor microenvironment. Existing clinical intervention strategies, such as target therapy and immunotherapy, have failed to achieve good therapeutic effects. In this article, single-cell transcriptome sequencing (scRNA-seq) data from six patients downloaded from the GEO database were adopted to describe the tumor microenvironment (TME) of ccRCC, including its T cells, tumor-associated macrophages (TAMs), endothelial cells (ECs), and cancer-associated fibroblasts (CAFs). Based on the differential typing of the TME, we identified tumor cell-specific regulatory programs that are mediated by three key transcription factors (TFs), whilst the TF EPAS1/HIF-2α was identified via drug virtual screening through our analysis of ccRCC's protein structure. Then, a combined deep graph neural network and machine learning algorithm were used to select anti-ccRCC compounds from bioactive compound libraries, including the FDA-approved drug library, natural product library, and human endogenous metabolite compound library. Finally, five compounds were obtained, including two FDA-approved drugs (flufenamic acid and fludarabine), one endogenous metabolite, one immunology/inflammation-related compound, and one inhibitor of DNA methyltransferase (N4-methylcytidine, a cytosine nucleoside analogue that, like zebularine, has the mechanism of inhibiting DNA methyltransferase). Based on the tumor microenvironment characteristics of ccRCC, five ccRCC-specific compounds were identified, which would give direction of the clinical treatment for ccRCC patients.

Hybrid hyperplane gradient learning algorithm for RBF neural network

Gradient-based algorithms are still popularly used for training radial basis function neural network (RBFNN). However, these algorithms may lead to the vanishing gradient (saddle point or local minimum) problem, which is one of the common problems that limit learning performance. To cope with this, a hybrid hyperplane gradient learning algorithm (HHGLA) is proposed to improve the learning performance of the RBFNN in this paper. First, a hyperplane gradient (HPG), based on a hyperspace constructed by a solution population (SP), is introduced. Thus, the search process can cross the surface of the cost function to avoid the appearance of vanishing gradient. Second, an adaptive learning rate is designed to restrict the search process into this hyperspace. Then, during the learning process, the hyperspace is constantly contracted to approximate the global optimal solution. Third, the convergence property of HHGLA-based RBFNN is thoroughly analyzed to ensure its application success. Finally, empirical comparisons of the proposed HHGLA-RBFNN are given to illustrate its superiority. The simulation results show the feasibility of HHGLA-RBFNN of accuracy.

Threshold learning algorithm for memristive neural network with binary switching behavior

On-chip learning is an effective method for adjusting artificial neural networks in neuromorphic computing systems by considering hardware intrinsic properties. However, it faces challenges due to hardware nonidealities, such as the nonlinearity of potentiation and depression and limitations on fine weight adjustment. In this study, we propose a threshold learning algorithm for a variation-tolerant ternary neural network in a memristor crossbar array. This algorithm utilizes two tightly separated resistance states in memristive devices to represent weight values. The high-resistance state (HRS) and low-resistance state (LRS) defined as read current of < 0.1 μA and > 1 μA, respectively, were successfully programmed in a 32 × 32 crossbar array, and exhibited half-normal distributions due to the programming method. To validate our approach experimentally, a 64 × 10 single-layer fully connected network were trained in the fabricated crossbar for an 8 × 8 MNIST dataset using the threshold learning algorithm, where the weight value is updated when a gradient determined by backpropagation exceeds a threshold value. Thanks to the large margin between the two states of the memristor, we observed only a 0.42% drop in classification accuracy compared to the baseline network results. The threshold learning algorithm is expected to alleviate the programming burden and be utilized in variation-tolerant neuromorphic architectures.

Understanding and Improving Optimization in Predictive Coding Networks

Backpropagation (BP), the standard learning algorithm for artificial neural networks, is often considered biologically implausible. In contrast, the standard learning algorithm for predictive coding (PC) models in neuroscience, known as the inference learning algorithm (IL), is a promising, bio-plausible alternative. However, several challenges and questions hinder IL's application to real-world problems. For example, IL is computationally demanding, and without memory-intensive optimizers like Adam, IL may converge to poor local minima. Moreover, although IL can reduce loss more quickly than BP, the reasons for these speedups or their robustness remains unclear. In this paper, we tackle these challenges by 1) altering the standard implementation of PC circuits to substantially reduce computation, 2) developing a novel optimizer that improves the convergence of IL without increasing memory usage, and 3) establishing theoretical results that help elucidate the conditions under which IL is sensitive to second and higher-order information.

Risk prediction and analysis of gallbladder polyps with deep neural network

The aim of this study is to analyze the risk factors associated with the development of adenomatous and malignant polyps in the gallbladder. Adenomatous polyps of the gallbladder are considered precancerous and have a high likelihood of progressing into malignancy. Preoperatively, distinguishing between benign gallbladder polyps, adenomatous polyps, and malignant polyps is challenging. Therefore, the objective is to develop a neural network model that utilizes these risk factors to accurately predict the nature of polyps. This predictive model can be employed to differentiate the nature of polyps before surgery, enhancing diagnostic accuracy. A retrospective study was done on patients who had cholecystectomy surgeries at the Department of Hepatobiliary Surgery of the Second People’s Hospital of Shenzhen between January 2017 and December 2022. The patients’ clinical characteristics, lab results, and ultrasonographic indices were examined. Using risk variables for the growth of adenomatous and malignant polyps in the gallbladder, a neural network model for predicting the kind of polyps will be created. A normalized confusion matrix, PR, and ROC curve were used to evaluate the performance of the model. In this comprehensive study, we meticulously analyzed a total of 287 cases of benign gallbladder polyps, 15 cases of adenomatous polyps, and 27 cases of malignant polyps. The data analysis revealed several significant findings. Specifically, hepatitis B core antibody (95% CI −0.237 to 0.061, p < 0.001), number of polyps (95% CI −0.214 to −0.052, p = 0.001), polyp size (95% CI 0.038 to 0.051, p < 0.001), wall thickness (95% CI 0.042 to 0.081, p < 0.001), and gallbladder size (95% CI 0.185 to 0.367, p < 0.001) emerged as independent predictors for gallbladder adenomatous polyps and malignant polyps. Based on these significant findings, we developed a predictive classification model for gallbladder polyps, represented as follows, Predictive classification model for GBPs = −0.149 * core antibody − 0.033 * number of polyps + 0.045 * polyp size + 0.061 * wall thickness + 0.276 * gallbladder size − 4.313. To assess the predictive efficiency of the model, we employed precision-recall (PR) and receiver operating characteristic (ROC) curves. The area under the curve (AUC) for the prediction model was 0.945 and 0.930, respectively, indicating excellent predictive capability. We determined that a polyp size of 10 mm served as the optimal cutoff value for diagnosing gallbladder adenoma, with a sensitivity of 81.5% and specificity of 60.0%. For the diagnosis of gallbladder cancer, the sensitivity and specificity were 81.5% and 92.5%, respectively. These findings highlight the potential of our predictive model and provide valuable insights into accurate diagnosis and risk assessment for gallbladder polyps. We identified several risk factors associated with the development of adenomatous and malignant polyps in the gallbladder, including hepatitis B core antibodies, polyp number, polyp size, wall thickness, and gallbladder size. To address the need for accurate prediction, we introduced a novel neural network learning algorithm. This algorithm utilizes the aforementioned risk factors to predict the nature of gallbladder polyps. By accurately identifying the nature of these polyps, our model can assist patients in making informed decisions regarding their treatment and management strategies. This innovative approach aims to improve patient outcomes and enhance the overall effectiveness of care.

Application of machine learning in medicine

In the past decades, the quantity and complexity of medical data have been increasing, such as medical images, genomics data, physiological signals and so on. These data contain a lot of valuable information, but traditional analysis methods and manual feature extraction have been unable to effectively deal with these large-scale and high-dimensional data. The rapid development of machine learning technology has brought new opportunities for medical research and clinical practice. Machine learning algorithms can learn and discover patterns, laws and prediction models from large-scale data, thus helping doctors and researchers to make more accurate and personalized diagnosis and treatment decisions. The application of machine learning in medicine has become a research field of great concern in recent years. This paper studies the application of machine learning in medicine, such as medical image diagnosis, genomics and drug discovery, and analyzes the relevant technical methods and ideas of machine learning in medicine, and analyzes the main algorithms and usage methods used in medical image diagnosis, such as convolution neural network and other deep learning algorithms. Through detailed analysis and research, it is found that machine learning provides a new method and tool, which can effectively process large-scale and complex medical data in medicine, bring more possibilities for medical diagnosis, treatment and research, and provide support for individualized medical care.

Spatiotemporal change analysis of LULC using remote sensing and CA-ANN approach in the Hodna basin, NE of Algeria

Spatiotemporal studies of land use/land cover change (LULCC) have become an important issue due to the rapid conversion of landscape and its impact on the environment and ecosystems. This study aims to identify the LULC change's impact on the Hodna basin (26,000 km2) from 2000 to 2020 using Landsat images data and forecast the possible change scenarios to the horizon of 2050. The Maximum Likelihood Supervised Classification (MLSC) was applied to Landsat data between 2000 and 2020 with a five-year interval. The confusion matrix of image classification indicates an overall accuracy of 85% and a kappa coefficients range between 0.82 and 0.96. To forecast the possible change scenarios of each class of LULC a transitional simulation using an artificial neural network (ANN) and machine learning algorithms. While, the cellular automata (CA) algorithm was used to simulate and predict the future LULC for the years 2030, 2040 and 2050 after validation of 2020 LULC map. The overall kappa calculated is 68.5% indicating a good accuracy of the validation process. The results show that for 2000–2020, agricultural lands decreased by 2.16% and sandy sails increased by 9.45%, indicating the expansion in the desertification phenomenon. Over the period 2000 to 2050, bare land decreased by 10% in 2015, recovered by 15% in 2020, and is projected to increase by 8% by 2050, reaching 73%; mixed and dense forest exhibited similar trends, growing from 0.7% in 2000 to 1.66% in 2050 with greater stability; built-up areas in the study zone surged by 150%, from 0.86% in 2000 to 2.5% in 2020, while clear forest remained relatively consistent, covering 4–6% of the surface; agricultural land diverged from bare land due to agricultural investment policies favoring new farms over bare land expansion. Urban sprawl is observed in the major cities like M'sila and Bordj Bou Arreridj and accelerated during the study period. Urban tissue evolved at the expense of agricultural land and focused mainly on slight slope areas and valleys.

Training multi-layer spiking neural networks with plastic synaptic weights and delays.

Spiking neural networks are usually considered as the third generation of neural networks, which hold the potential of ultra-low power consumption on corresponding hardware platforms and are very suitable for temporal information processing. However, how to efficiently train the spiking neural networks remains an open question, and most existing learning methods only consider the plasticity of synaptic weights. In this paper, we proposed a new supervised learning algorithm for multiple-layer spiking neural networks based on the typical SpikeProp method. In the proposed method, both the synaptic weights and delays are considered as adjustable parameters to improve both the biological plausibility and the learning performance. In addition, the proposed method inherits the advantages of SpikeProp, which can make full use of the temporal information of spikes. Various experiments are conducted to verify the performance of the proposed method, and the results demonstrate that the proposed method achieves a competitive learning performance compared with the existing related works. Finally, the differences between the proposed method and the existing mainstream multi-layer training algorithms are discussed.

Neural Network Algorithm with Reinforcement Learning for Microgrid Techno-Economic Optimization

Hybrid energy systems (HESs) are gaining prominence as a practical solution for powering remote and rural areas, overcoming limitations of conventional energy generation methods, and offering a blend of technical and economic benefits. This study focuses on optimizing the sizes of an autonomous microgrid/HES in the Kingdom of Saudi Arabia, incorporating solar photovoltaic energy, wind turbine generators, batteries, and a diesel generator. The innovative reinforcement learning neural network algorithm (RLNNA) is applied to minimize the annualized system cost (ASC) and enhance system reliability, utilizing hourly wind speed, solar irradiance, and load behavior data throughout the year. This study validates RLNNA against five other metaheuristic/soft-computing approaches, demonstrating RLNNA’s superior performance in achieving the lowest ASC at USD 1,219,744. This outperforms SDO and PSO, which yield an ASC of USD 1,222,098.2, and MRFO, resulting in an ASC of USD 1,222,098.4, while maintaining a loss of power supply probability (LPSP) of 0%. RLNNA exhibits faster convergence to the global solution than other algorithms, including PSO, MRFO, and SDO, while MRFO, PSO, and SDO show the ability to converge to the optimal global solution. This study concludes by emphasizing RLNNA’s effectiveness in optimizing HES sizing, contributing valuable insights for off-grid energy systems in remote regions.

Automatic fabrication system of optical micro-nanofiber based on deep learning

&lt;sec&gt;The wide range, high precision, and dynamic real-time measurement of micro-/nanofiber diameter are crucial for achieving low loss transmission and controlling dispersion in the preparation process of micro-nanofiber. In view of the problems of small diameter regulation range, complex operation and long-time consumption of the existing preparation methods, the automatic detection system of micro-nanofiber is realized based on deep learning neural network algorithm in this work. The image segmentation method in computer vision is used to make high-quality multi-scale micro-/nanofiber datasets, and the improved YOLOv8-FD (You Only Look Once version 8-Fiber Detection) algorithm based on small target detection is used to automatically detect the diameter of micro-nanofiber.&lt;/sec&gt;&lt;sec&gt;Through image segmentation and identification of the target of single pixel size in the microscopic image, the diameter detection of micro-nanofiber is finally realized. In this process, the real-time diameter of micro-nanofiber is obtained through image information, and then the micro-nanofiber small target is accurately segmented to achieve the precise detection of mAP&lt;sup&gt;IoU=50&lt;/sup&gt; = 0.995 and mAP&lt;sup&gt;IoU=50—95&lt;/sup&gt; = 0.765 on the micro-nanofiber multi-scale target dataset with extremely high accuracy. The algorithm-based construction of a high-precision micro-nanofiber automatic preparation system enables real-time accurate segmentation of fiber edges, calculation of fiber diameter, and feedback to the control system for achieving automated preparation of fibers with arbitrary diameters. Additionally, it facilitates the detection of micro-nanofiber in a range from 462 nm to 125 μm. The average response time for reasoning is 9.6 ms, and the detection error is kept below 2.95%.&lt;/sec&gt;&lt;sec&gt;In addition, compared with other micro-/nanofiber diameter detection methods based on optical imaging and mode cutoff, this method shows advantages of high precision, high speed and arbitrary diameter preparation for diameter detection based on deep learning neural networks. The system is very suitable for high-precision real-time measurement and automatic and accurate preparation of micro-/nanofibers, thereby providing a novel method of developing micro-/nanofiber devices with low-loss transmission and adjustable dispersion. All the data presented in this paper are openly available at &lt;ext-link ext-link-type="uri" xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="https://doi.org/10.57760/sciencedb.j00213.00021"&gt;https://doi.org/10.57760/sciencedb.j00213.00021&lt;/ext-link&gt;.&lt;/sec&gt;

IoT-based Deep Learning Neural Network (DLNN) algorithm for voltage stability control and monitoring of solar power generation

Today, Solar Photovoltaic (SPV) energy, an advancing and attractive clean technology with zero carbon emissions, is widely used. It is crucial to pay serious attention to the maintenance and application of Solar Power Generation (SPG) to harness it effectively. The design was more costly, and the automatic monitoring is not precise. The main objective of the work related to designed and built up the Internet of Things (IoT) platform to monitor the SPV Power Plants (SPVPP) to solve the issue. IoT platform designing and Data Analytics (DA) are the two phases of the proposed methodology. For building the IoT device in the IoT platform designing phase, diverse lower-cost sensors with higher end-to-end delivery ratio, higher network lifetime, throughput, residual energy, and better energy consumption are considered. Then, Sigfox communication technology is employed at the Low-Power Wireless Area Network (LPWAN) communication layer for lower-cost communication. Therefore, in the DA phase, the sensor monitored values are evaluated. In the analysis phase, which is the most significant part of the work, the input data are first pre-processed to avoid errors. Next, to monitor the Energy Loss (EL), the fault, and Potential Energy (PE), the solar features are extracted as of the pre-processed data. The significance of utilizing the Transformation Search centered Seagull Optimization (TSSO) algorithm, the significant features are chosen as of the extracted features. Therefore, the computational time of the solar monitoring has been decreased by the Feature Selection (FS). Next, the features are input into the Gaussian Kernelized Deep Learning Neural Network (GKDLNN) algorithm, which predicts the faults, PE, and EL. In the experimental evaluation, solar generation is assessed based on Wind Speed (WS), temperature, time, and Global Solar Radiation (GSR). The systems are satisfactory and produce more power during the time interval from 12:00 PM to 1:00 PM. The performance of the proposed method is evaluated based on performance metrics and compared with existing research techniques. When compared to these techniques, the proposed framework achieves superior results with improved precision, accuracy, F-measure, and recall.

Research on Shovel-Force Prediction and Power-Matching Optimization of a Large-Tonnage Electric Wheel Loader

Nowadays, rapid development has been achieved with respect to the electric wheel loader (EWL). The operational efficiency of EWLs is affected by many factors; especially, shovel force is a very important factor. For large-tonnage EWLs, when employing empirical, formula-based methods to predict shovel force, the generated errors are significant, with errors frequently reaching levels of up to 30%. To solve this problem, a method, based on the discrete element method (DEM), to predict shovel force is put forward in this paper. The material parameters are calibrated by a backpropagation (BP) neural network learning algorithm (NNLA). The material model is inputted into multi-body-dynamics software. A simulation model to accurately predict the shovel force is created. The error between the test results and the simulation results is 7.8%, demonstrating a high level of consistency. To validate the reliability of this method, the 35-ton EWL is taken as an example for research, and the straight-line driving test and the power-matching test are conducted. While ensuring the operational efficiency of the EWLs, the power loss is also a crucial consideration. The drastic changes in shovel force often result in front-tire slippage of the EWLs. To minimize wheel slippage during the shoveling section, the matching of the electric motor was optimized. In summary, material parameters were calibrated using a combined method of BP NNLA to predicate shovel force of a large-tonnage EWL. Additionally, the power matching of the EWL has been optimized to accord with the shoveling section of the device.

SOINN Intrusion Detection Model Based on Three-Way Attribute Reduction

With a large number of intrusion detection datasets and high feature dimensionality, the emergent nature of new attack types makes it impossible to collect network traffic data all at once. The modified three-way attribute reduction method is combined with a Self-Organizing Incremental learning Neural Network (SOINN) algorithm to propose a self-organizing incremental neural network intrusion detection model based on three-way attribute reduction. Attribute importance is used to perform attribute reduction, and the data after attribute reduction are fed into a self-organized incremental learning neural network algorithm, which generalizes the topology of the original data through self-organized competitive learning. When the streaming data are transferred into the model, the inter-class insertion or node fusion operation is performed by comparing the inter-node distance and similarity threshold to achieve incremental learning of the model streaming data. The inter-node distance value is introduced into the weight update formulation to replace the traditional learning rate and to optimize the topological structure adjustment operation. The experimental results show that T-SOINN achieves high precision and recall when processing intrusion detection data.

Identification of structured nonlinear state–space models for hysteretic systems using neural network hysteresis operators

Hysteretic system behavior is ubiquitous in science and engineering fields including measurement systems and applications. In this paper, we put forth a nonlinear state–space system identification method that combines the state–space equations to capture the system dynamics with a compact and exact artificial neural network (ANN) representation of the classical Prandtl–Ishlinskii (PI) hysteresis. These ANN representations called PI hysteresis operator neurons employ recurrent ANNs with classical activation functions, and thus can be trained with classical neural network learning algorithms. The structured nonlinear state–space model class proposed in this paper, for the first time, offers a flexible interconnection of PI hysteresis operators with a linear state–space model through a linear fractional representation. This results in a comprehensive and flexible model structure. The performance is validated both on numerical simulation and on measurement data.

Autonomous navigation method of jujube catch-and-shake harvesting robot based on convolutional neural networks

Autonomous navigation of agricultural devices plays an indispensable step for the realization of various field intelligent operation tasks. However, the accurate extraction of navigation path is still a challenging task due to the complex orchard environment with many interference factors. To cope with the challenge, a low-cost vision system based on convolutional neural network and machine learning algorithm is proposed to obtain accurate navigation path for autonomous path detection of catch-and-shake harvesting robot under winter jujube orchard. In the vision system, firstly, an improved light-weight YOLOX-Nano architecture with some tricks including CSP attention block, SPPCSPC-F and ASFF is proposed to detect root points of jujube tree. Results show that the mAP of our model is 84.08 % and model size is only 12.30 MB, which can meet the requirements of embedded deployment. Secondly, a K-means clustering algorithm is applied to divide root points into two classes which belong to left and right tree row. Finally, the navigation line can be determined by geometric relations and least squares. The experimental results showed that the navigation centerline mean pixel errors and average heading angle deviation are 4.49 pixels and 2.55°, respectively. Therefore, this method can provide a new theoretical basis and technical reference for autonomous navigation of winter jujube catch-and-shake harvesting robots.

Human Activity Recognition Using CNN and Lstm Deep Learning Algorithms

Human Activity Recognition recognizes and classifies the activities performed by the users or people based on the data collected from the sensors of special devices such as smart-watches, smartphones etc. It has become easy to collect a huge amount of data from inertial sensors that are embedded in wearable devices. An accelerometer and gyroscope sensors are most commonly used inertial sensors. There are various already available datasets, in our paper, we are using the Wireless Sensor Data Mining dataset which contains 1,098,207 data of 6 physical activities performed. In this paper, the activities we aim to classify are walking, jogging, going up and downstairs, standing, and sitting. There are various algorithms applied on the various datasets. In our paper, we use Convolutional Neural Network and Long Short-Term Memory deep learning algorithm on the data set, we split the data into training data [80%] and testing data [20%]. By using a confusion matrix, we recognize and classify the activities performed using maximum accuracy.

Learning algorithms for oscillatory neural networks as associative memory for pattern recognition

Alternative paradigms to the von Neumann computing scheme are currently arousing huge interest. Oscillatory neural networks (ONNs) using emerging phase-change materials like VO2 constitute an energy-efficient, massively parallel, brain-inspired, in-memory computing approach. The encoding of information in the phase pattern of frequency-locked, weakly coupled oscillators makes it possible to exploit their rich non-linear dynamics and their synchronization phenomena for computing. A single fully connected ONN layer can implement an auto-associative memory comparable to that of a Hopfield network, hence Hebbian learning rule is the most widely adopted method for configuring ONNs for such applications, despite its well-known limitations. An extensive amount of literature is available about learning in Hopfield networks, with information regarding many different learning algorithms that perform better than the Hebbian rule. However, not all of these algorithms are useful for ONN training due to the constraints imposed by their physical implementation. This paper evaluates different learning methods with respect to their suitability for ONNs. It proposes a new approach, which is compared against previous works. The proposed method has been shown to produce competitive results in terms of pattern recognition accuracy with reduced precision in synaptic weights, and to be suitable for online learning.