Data mining is the process to predict future trends. Data mining involving searching of patterns whose general purpose is to extract information using intelligent methods from a set of data and information turn it int...
详细信息
ISBN:
(数字)9781728183794
ISBN:
(纸本)9781728183800
Data mining is the process to predict future trends. Data mining involving searching of patterns whose general purpose is to extract information using intelligent methods from a set of data and information turn it into an understandable structure for later use. Data mining using IoT becomes one of the leading providers applicable in several areas. The method of extracting usable data from larger set of arbitrary raw data is called data mining. This includes the analysis of sample data in large amounts of data with one or more programs. Concentrate on large amounts of data and databases to analyze, we use predictive analysis in the field of medicine. In this study, we examine the performance of algorithms along with the use of IoT that analyses large amounts of scattered information to make sense of it and turn it into knowledge. For this we used feature selection and classification algorithm on hepatitis data to get best result with minimum error rate and compare feature selection with classification algorithm so that to check which method is best according to results.
This research examines the effectiveness of various machine learning models in accurately classifying the intent of agricultural queries using fundamental machine learning techniques. The study analyzes two curated da...
详细信息
ISBN:
(数字)9798350380040
ISBN:
(纸本)9798350380057
This research examines the effectiveness of various machine learning models in accurately classifying the intent of agricultural queries using fundamental machine learning techniques. The study analyzes two curated datasets: the QUERY dataset, containing 2,853 queries and their corresponding intents, and the ANSWERS dataset, which includes crops and their associated intents, covering 21 intent categories across 30 different crops. The main objective is to evaluate the performance of traditional machine learning models on this dataset to enable accurate intent classification, facilitating responsive interactions via the QA system, which delivers data-driven answers to farmers' queries. Notably, the Support Vector Classifier demonstrated the best performance with an accuracy of 92 % , closely followed by the Linear Discriminant Analysis Classifier. Using the Support Vector Classifier, the QA system efficiently retrieves responses from the ANSWERS dataset. This paper provides a comprehensive analysis of the performance of various algorithms for the question-answering (QA) task in the agricultural domain. The evaluation is based on pre-modelled classification reports from Sklearn's metrics package, ensuring rigorous analysis and detailed insights into the effectiveness of each approach.
We present a novel scheme, which enables the receiver to automatically identify the channel noise type in image communication. The method is based on embedding the image histogram, and deriving its specific statistics...
详细信息
We present a novel scheme, which enables the receiver to automatically identify the channel noise type in image communication. The method is based on embedding the image histogram, and deriving its specific statistics as descriptive features of the original image, through a robust watermarking algorithm within itself at the transmitter and extracting the hidden data at the receiver, for comparing with the corresponding features of the noise-distorted image. Then, by using a classifier, we are able to distinguish the channel noise type. Implementation results prove the efficiency of our proposed system in capably recognizing the noise type for common image transmission noise varieties.
Aiming at the problems of high computational complexity and insufficient accuracy of traditional deep learning image classification algorithms, a novel network structure called DARN (Dual Attention ResNet) is designed...
详细信息
ISBN:
(数字)9798331528911
ISBN:
(纸本)9798331528928
Aiming at the problems of high computational complexity and insufficient accuracy of traditional deep learning image classification algorithms, a novel network structure called DARN (Dual Attention ResNet) is designed. This method embeds channel attention and spatial attention modules in the residual blocks, optimizes the training strategy, and reduces the algorithm complexity. On the ImageNet dataset, the classification accuracy reaches 78.6%, which is 2.3% higher than the original ResNet50, with a 20.5% reduction in parameters and a 31% increase in inference speed, achieving a balance between model performance and computational efficiency.
Two classification algorithms that rely on both spectral and textural information are presented and compared. The first is a standard maximum-likelihood classification procedure with a texture "band" added t...
详细信息
Two classification algorithms that rely on both spectral and textural information are presented and compared. The first is a standard maximum-likelihood classification procedure with a texture "band" added to the spectral band set. The second is a pattern matching algorithm which integrates the spectral and spatial characteristics of the data in recognizing a user-specified training pattern. The pattern matching algorithm proved to be the most effective procedure of those compared. classification results from both methods are compared with each other and with a purely spectral classification using a maximum-likelihood classifier.< >
In the past decade, the accelerometer has been used to enable activity recognition in different application domains. In recent years, the accelerometer in a smartphone is also being applied to provide unobtrusive move...
详细信息
In the past decade, the accelerometer has been used to enable activity recognition in different application domains. In recent years, the accelerometer in a smartphone is also being applied to provide unobtrusive movement recognition. Majority of the existing investigations requires the orientation of the sensor device to be fixed. By applying the orientation-independent approach, proposed by Mizell, this requirement may be relaxed. In this paper, we compare the recognition accuracy using classification algorithms built from raw and orientation-independent acceleration data. The evaluations, based on acceleration data collected from five users, have shown that the application of the orientation-independent approach achieves accuracy up to 88 %. The trade-off of relaxing the requirement of fixed-orientation is around 5-6 %.
Internet of Things (IoT) is state of the art technology of internet network that enables devices in our life (e.g., camera, and cars) to connect internet network and exchange data between them. Data communication betw...
详细信息
Internet of Things (IoT) is state of the art technology of internet network that enables devices in our life (e.g., camera, and cars) to connect internet network and exchange data between them. Data communication between these devices exposed to different attacks (e.g., Denial of Service (DoS), buffer overflow, and probing attack). So we need secure and good performance techniques to detect these attacks. In this paper, we use Artificial Neural Networks (ANN) classification algorithms to detect several attacks which IoT communication is exposed. These algorithms are all learning Vectors Quantization (LVQ's) versions, Radial basis function (RBN) and Multilayer Perceptron (MLP). In this paper, to conduct our experimental results, we use KDD CUP 99 Dataset that contains 494020 instances for different attacks. Our results showed that LVQ2.1 archives best classification accuracy (97.44%) than other versions during 4.56 second. On the other hand, MLP achieves best accuracy (99.86%) than LVQ2.1 and RBF. Unfortunately, time taken by MLP is significantly low which takes 28811 seconds.
Ancillary Services Market in Italy is knowing a period of reformation: Distributed Generators and loads are now enabled for the provision of dispatching resources. This calls for the development of fast, cheap and rel...
详细信息
Ancillary Services Market in Italy is knowing a period of reformation: Distributed Generators and loads are now enabled for the provision of dispatching resources. This calls for the development of fast, cheap and reliable tools to be used by operators with poor awareness about energy markets. In this paper a first approach to the problem is proposed: Machine Learning-based classification models are developed and tested over a set of pre-processed market data. Then, a selected model based on Decision Trees is further elaborated to test its sensitivity with respect to hyperparameters tuning and learning techniques. Results highlighted the possibility to exploit this kind of models to integrate market-based logic in the control and automation of Distributed Generators and microgrids.
In this paper, a novel packet classification scheme optimized for multi-core network processors is proposed. The algorithm, Explicit Cuttings (ExpCuts), adopts a hierarchical space aggregation technique to significant...
详细信息
In this paper, a novel packet classification scheme optimized for multi-core network processors is proposed. The algorithm, Explicit Cuttings (ExpCuts), adopts a hierarchical space aggregation technique to significantly reduce the memory usage. Consequently, without burst of memory usages, the time-consuming linear search in the conventional decision-tree based packet classification algorithms is eliminated, and an explicit worst-case search time is achieved. To evaluate the performance of ExpCuts, we implement the algorithm, as well as HiCuts and HSM, on the Intel IXP2850 network processor. Experimental results show that ExpCuts outperforms the existing best- known algorithms in terms of memory usage and classification speed.
CNES (French Space Agency) has developed a research program related to SPOT imagery to deal with cartography topics. Many studies, conducted with different laboratories, are intended to work on remote sensing data. Th...
详细信息
CNES (French Space Agency) has developed a research program related to SPOT imagery to deal with cartography topics. Many studies, conducted with different laboratories, are intended to work on remote sensing data. The main purpose of the present research is information extraction (network extraction, urban area extraction, segmentation, etc.) One of these studies, made in collaboration with INRIA Sophia Antipolis, intends to classify remote sensing images using MRF (Markov random field) modelization. The paper presents an experiment, conducted by GEOSYS, on crop surveys. An evaluation of the MRF based algorithms is proposed to estimate the results in a supervised context, in order to validate this new approach. A comparison between the proposed methods and standard classification techniques have been done on multispectral SPOT data (XS1, XS2, XS3).
暂无评论