Affected by haze, images often face color distortion, resolution reduction and other image quality degradation problems. the existing dehazing methods based on convolutional neural network(CNN) often perform well on l...
详细信息
In the initial stage of the COVID-19 pandemic, assessing public opinion is the most critical problem to combat the epidemic by enforcing a national lockdown, rendering health services, and presenting quarantine proced...
详细信息
Cognitive impairment detection is challenging because it primarily depends on advanced neuroimaging tests, which are not readily available in small cities and villages in India and many developing countries. Artificia...
详细信息
ISBN:
(纸本)9783031451690;9783031451706
Cognitive impairment detection is challenging because it primarily depends on advanced neuroimaging tests, which are not readily available in small cities and villages in India and many developing countries. Artificial Intelligence (AI)-based systems can be used to assist clinical decision-making, but it requires advanced tests and a large amount of data to achieve reasonable accuracy. In this work, we have developed an explainable decision-tree-based detection model, which serves as a powerful first-level screening on a small basic set of cognitive tests. this minimum set of features is obtained through an ablation study. the Alzheimer's Disease Neuroimaging Initiative (ADNI) archive provided the data for this study. We obtained 93.10% accuracy for three classes: Cognitive Normal (CN), Mild Cognitive Impairment (MCI), Alzheimer's Disease (AD), and 78.74% accuracy for five classes: CN, Significant Memory Complaints (SMC), Early-MCI (EMCI), Late-MCI (LMCI), AD using Extreme Gradient Boosting (XGBoost) which is comparable to the accuracy of state-of-the-art methods which use a more sophisticated and expensive test like imaging. Current research pays little attention to explainability and is primarily concerned with enhancing the performance of deep learning and machinelearning models. Consequently, clinicians find it challenging to interpret these intricate models. Withthe use of Tree Shapley Additive Explanations (TreeSHAP) values and Local Interpretable Model-agnostic Explanations (LIME), this work intends to give both global and local interpretability respectively. Moreover, we highlight the use of top-2 metrics (Accuracy, Precision, and Recall), which significantly improves corner cases and helps the clinician streamline diagnosis.
A Real time gesture recognition system employs and leverages on the concepts of machinelearning to interpret and respond promptly to the user gestures, significantly enhancing the Human Computer Interaction. Instead ...
详细信息
ISBN:
(数字)9798331505790
ISBN:
(纸本)9798331505806
A Real time gesture recognition system employs and leverages on the concepts of machinelearning to interpret and respond promptly to the user gestures, significantly enhancing the Human Computer Interaction. Instead of analyzing one frame and predicting the gesture, we use a set of frames to determine the action. Media Pipe is used for the process for key point extraction of the hand joints, which provides comprehensive data for the analysis. For the prediction in real time, the Long Short Term Memory (LSTM) model is trained using TensorFlow. Keras is used for building and training neural networks in Python, alongside TensorFlow to streamline the model development process. this system aims to have seamless interaction between the machinelearning model and the gesture data extraction to enhance the user experience and opens up new possibilities for intuitive and natural communication with computers.
the outbreak of COVID-19 has lasted for two years. the rapid spread and fatal variability of COVID-19 pose a great threat to human survival. Today, the existing high-tech medical technology has not found a direct spec...
详细信息
ISBN:
(纸本)9781450397636
the outbreak of COVID-19 has lasted for two years. the rapid spread and fatal variability of COVID-19 pose a great threat to human survival. Today, the existing high-tech medical technology has not found a direct specific drug. therefore, efficient diagnostic techniques and methods play a key role in controlling the spread of COVID-19 and managing patients' conditions. Deep learning technology can learn implicit samples of data. this paper mainly studies the nonlinear relationship between the serum Raman spectrum data of new crown and healthy people by using convolutional neural network, effectively expand the samples of training set by using data enhancement method, standardize the spectral data, smooth denoising by savitzky Golay method, and construct the prediction model based on convolutional neural network after principal component analysis. Compared with other traditional machinelearning algorithms, the features extracted by convolution neural network through convolution layer, batch standardization layer and pooling layer are more comprehensive, which can effectively improve the accuracy and speed of COVID-19 recognition and classification. the experimental results show that convolution neural network has a higher screening accuracy for COVID-19, and the accuracy rate is 98.39%, It is proved that Raman spectroscopy combined with deep learning is effective and feasible in screening COVID-19.
the proceedings contain 17 papers. the topics discussed include: can GPT-4 support analysis of textual data in tasks requiring highly specialized domain expertise?;towards meaningful paragraph embeddings for data-scar...
the proceedings contain 17 papers. the topics discussed include: can GPT-4 support analysis of textual data in tasks requiring highly specialized domain expertise?;towards meaningful paragraph embeddings for data-scarce domains: a case study in the legal domain;enhancing pre-trained language models with sentence position embeddings for rhetorical roles recognition in legal opinions;bridging the gap: mapping layperson narratives to legal issues with language models;taking the law more seriously by investigating design choices in machinelearning prediction research;semantic extraction of key figures and their properties from tax legal texts using neural models;contrast is all you need;organizing the unorganized: a novel approach for transferring a taxonomy of labels into flat-labeled document collections;and chasing the invisible in the grammar of repetitions: a network analysis approach to fiscal state aids.
Currently, most of the people are stating their opinion through social media. Public opinion stands an important one while reviewing any product, Movie etc. the field of opinion mining has turn out to be the most sign...
详细信息
Ensuring the security of information systems in the modern technology era is a critical requirement due to the increasing complexity and volume of malicious codes and threats. However, traditional signature-based appr...
详细信息
the present article applies unsupervised machinelearning approach to identify patterns of dependence in data of biomarkers, cognitive and demographic characteristics useful for the diagnosis and treatment planning of...
详细信息
the proceedings contain 31 papers. the special focus in this conference is on machinelearning and datamining in patternrecognition. the topics include: An information retrieval approach for finding dependent subspa...
ISBN:
(纸本)9783319624150
the proceedings contain 31 papers. the special focus in this conference is on machinelearning and datamining in patternrecognition. the topics include: An information retrieval approach for finding dependent subspaces of multiple views;predicting target events in industrial domains;importance of recommendation policy space in addressing click sparsity in personalized advertisement display;global flow and temporal-shape descriptors for human action recognition from 3D reconstruction data;reverse engineering gene regulatory networks using sampling and boosting techniques;detecting large concept extensions for conceptual analysis;qualitative and descriptive topic extraction from movie reviews using LDA;detecting relative anomaly;optimization for large-scale machinelearning with distributed features and observations;a scalable and noise-resistant closed contiguous sequential patterns mining algorithm;sparse dynamic time warping;over-fitting in model selection with Gaussian process regression;machinelearning-as-a-service and its application to medical informatics;anomaly detection from kepler satellite time-series data;prediction of insurance claim severity loss using regression models;a spectral clustering method for large-scale geostatistical datasets;vulnerability of deep reinforcement learning to policy induction attacks;mobile robot localization via machinelearning;an analysis of the application of simplified silhouette to the evaluation of k-means clustering validity;summarization-guided greedy optimization of machinelearning model;clustering aided support vector machines;mining player ranking dynamics in team sports and personalized visualization based upon wavelet transform for interactive software customization.
暂无评论