Image denoising based on deep learning has made more extensive development by using a large amount of data for network training, however, it is difficult to obtain clean images without noise in actual scenarios, leadi...
详细信息
machinelearning is coming everywhere, and currently in every field, it is contributing in terms of different applications. machinelearning is a type of artificial intelligence (AI), which enables a program to learn,...
详细信息
An important field within Human Activity recognition is the evaluation of disease and patient recovery through the assessment of gait patterns. Clustering has been used as a data-mining technique to find the prior pat...
详细信息
ISBN:
(纸本)9798350338416
An important field within Human Activity recognition is the evaluation of disease and patient recovery through the assessment of gait patterns. Clustering has been used as a data-mining technique to find the prior patterns in subjects' gait patterns. Previous studies have shown the discriminative power of gait clustering on biometrics, and the ability to detect abnormal gait patterns and gait pathology. Previous techniques have relied on expensive machinery and closed environments for gait pattern extraction and/or simplistic featured approaches to clustering. Geometric time series clustering has developed in other fields as a method for incorporating the information from an entire time series sequence and comparing sequences with temporal distortions. We present a method for geometric gait clustering using accelerometer data from wearable sensors. Our methods include an approach to gait cycle averaging and a two-way clustering method for assessing the similarity of biometrics within gait cycle clusters. Our results demonstrate that our methods have significant discriminative efficacy for biometrics and may be a useful analytical tool for gait pathology.
End-to-end text image translation (TIT), which aims at translating the source language embedded in images to the target language, has attracted intensive attention in recent research. However, data sparsity limits the...
详细信息
ISBN:
(数字)9781665490627
ISBN:
(纸本)9781665490627
End-to-end text image translation (TIT), which aims at translating the source language embedded in images to the target language, has attracted intensive attention in recent research. However, data sparsity limits the performance of end-to-end text image translation. Multi-task learning is a non-trivial way to alleviate this problem via exploring knowledge from complementary related tasks. In this paper, we propose a novel text translation enhanced text image translation, which trains the end-to-end model with text translation as an auxiliary task. By sharing model parameters and multi-task training, our model is able to take full advantage of easily-available large-scale text parallel corpus. Extensive experimental results show our proposed method outperforms existing end-to-end methods, and the joint multi-task learning with both text translation and recognition tasks achieves better results, proving translation and recognition auxiliary tasks are complementary. (1)
Multi-view clustering has always been a widely concerned issue due to its wide range of applications. Since real-world datasets are usually very large, the clustering problem for large-scale multi-view datasets has al...
详细信息
ISBN:
(纸本)9789819984619;9789819984626
Multi-view clustering has always been a widely concerned issue due to its wide range of applications. Since real-world datasets are usually very large, the clustering problem for large-scale multi-view datasets has always been a research hotspot. Most of the existing methods to solve the problem of large-scale multi-view data usually include several independent steps, namely anchor point generation, graph construction, and clustering result generation, which generate the inflexibility anchor points, and the process of obtaining the cluster indicating matrix and graph constructing are separating from each other, which leads to suboptimal results. therefore, to address these issues, a one-step multi-view subspace clustering model based on orthogonal matrix factorization with consensus graph learning(CGLMVC) is proposed. Specifically, our method puts anchor point learning, graph construction, and clustering result generation into a unified learning framework, these three processes are learned adaptively to boost each other which can obtain flexible anchor representation and improve the clustering quality. In addition, there is no need for post-processing steps. this method also proposes an alternate optimization algorithm for convergence results, which is proved to have linear time complexity. Experiments on several real world large-scale multi-view datasets demonstrate its efficiency and scalability.
To study land use pattern with satellite images, there are many procedures for image processing. Like feature extraction from many images, image enhancement by removing noise, classification of images, and ultimately ...
详细信息
Human Activity recognition(HAR) is used in many applications, such as surveillance, anti-terrorists, anti-crime securities, medical, life logging, and assistance. Besides its positive effect on well-being, the recogni...
详细信息
Skeleton-based action/gesture recognition has already witnessed excellent progress on processing large-scale, laboratory-based datasets with pre-defined skeleton joint topology. However, it's still an unsolved tas...
详细信息
ISBN:
(数字)9781665490627
ISBN:
(纸本)9781665490627
Skeleton-based action/gesture recognition has already witnessed excellent progress on processing large-scale, laboratory-based datasets with pre-defined skeleton joint topology. However, it's still an unsolved task when it comes to real-world scenarios with practical limitations such as small-scaled dataset sizes, few-labeled samples, and various skeleton topologies. In this paper, we work on the recognition of micro-gestures, which are subtle body gestures collected in real-world scenarios. Specifically, we utilize contrastive learning to heritage the knowledge from known large-scale datasets for enhancing the learning on fewer samples of micro-gestures. To overcome the gap caused by various domain distributions and structure topologies between the datasets, we compute skeleton representations from augmented sequences via momentum-based efficient and scalable encoders as additional inductive priors. Importantly, we propose an effective dense-graph based unsupervised architecture that resorts to a queue-based dictionary to store positive and negative keys for better contrast with queries to learn substantially efficient and discriminant patterns in the feature space. Together with cross-dataset experimental results show that our model significantly improves the accuracies on two micro-gesture datasets, SMG by 7.4% and iMiGUE by 18.41% advocating its superiority.
this research is focused on enhancing the efficacy of direct marketing for bank deposit products through datamining analysis. the primary challenge lies in accurately predicting and targeting prospective customers. T...
详细信息
Domain generalization methods aim to learn models robust to domain shift withdata from a limited number of source domains and without access to target domain samples during training. Popular domain alignment methods ...
详细信息
ISBN:
(数字)9781665490627
ISBN:
(纸本)9781665490627
Domain generalization methods aim to learn models robust to domain shift withdata from a limited number of source domains and without access to target domain samples during training. Popular domain alignment methods for domain generalization seek to extract domain-invariant features by minimizing the discrepancy between feature distributions across all domains, disregarding inter-domain relationships. In this paper, we instead propose a novel representation learning methodology that selectively enforces prediction consistency between source domains estimated to be closely-related. Specifically, we hypothesize that domains share different class-informative representations, so instead of aligning all domains which can cause negative transfer, we only regularize the discrepancy between closely-related domains. We apply our method to time-series classification tasks and conduct comprehensive experiments on three public real-world datasets. Our method significantly improves over the baseline and achieves better or competitive performance in comparison with state-of-the-art methods in terms of both accuracy and model calibration.
暂无评论