Glaucoma is currently one of the most significant causes of permanent blindness. Fundus imaging is the most popular glaucoma screening method because of the compromises it has to make in terms of portability, size, an...
详细信息
Glaucoma is currently one of the most significant causes of permanent blindness. Fundus imaging is the most popular glaucoma screening method because of the compromises it has to make in terms of portability, size, and cost. In recent years, convolution neural networks (CNNs) have revolutionized computer vision. Convolution is a "local" CNN technique that is only applicable to a small region surrounding an image. Vision Transformers (ViT) use self-attention, which is a "global" activity since it collects information from the entire image. As a result, the ViT can successfully gather distant semantic relevance from an image. This study examined several optimizers, including Adamax, SGD, RMSprop, Adadelta, Adafactor, Nadam, and Adagrad. With 1750 Healthy and Glaucoma images in the IEEE fundus image dataset and 4800 healthy and glaucoma images in the LAG fundus image dataset, we trained and tested the ViT model on these datasets. Additionally, the datasets underwent image scaling, auto-rotation, and auto-contrast adjustment via adaptive equalization during preprocessing. The results demonstrated that preparing the provided dataset with various optimizers improved accuracy and other performance metrics. Additionally, according to the results, the Nadam Optimizer improved accuracy in the adaptive equalized preprocessing of the IEEE dataset by up to 97.8% and in the adaptive equalized preprocessing of the LAG dataset by up to 92%, both of which were followed by auto rotation and image resizing processes. In addition to integrating our vision transformer model with the shift tokenization model, we also combined ViT with a hybrid model that consisted of six different models, including SVM, Gaussian NB, Bernoulli NB, Decision Tree, KNN, and Random Forest, based on which optimizer was the most successful for each dataset. Empirical results show that the SVM Model worked well and improved accuracy by up to 93% with precision of up to 94% in the adaptive equalization preprocess
Rapid adaptation to the environment is the long-term task of reinforcement learning. However, reinforcement learning faces great challenges in dynamic environments, especially with continuous state-action spaces. In t...
详细信息
Community search over bipartite graphs is a critical issue that has garnered considerable interest. Nevertheless, existing research often neglects the importance of vertex weights in community formation, which leads t...
详细信息
InAs/AlAs superlattice structures have significant potential for application in low-noise avalanche *** their performance in practical applications linked to the fundamental physical properties of carrier relaxation t...
详细信息
InAs/AlAs superlattice structures have significant potential for application in low-noise avalanche *** their performance in practical applications linked to the fundamental physical properties of carrier relaxation time,this study investigated the carrier relaxation times of InAs/AlAs superlattices across various monolayers,temperatures,and carrier *** investigation indicated that relaxation times span several tens of picoseconds,confirming that high-quality interfaces do not significantly reduce relaxation times in the way defect states ***,our study demonstrates that adjustments to the superlattice period can effectively modulate both the bandgap and carrier relaxation times,potentially impacting the performance of avalanche photodiodes by altering the electron-phonon interaction pathways and bandgap *** established that lower temperatures contribute to an increase in the bandgap and the suppression of high-frequency optical phonon vibrations,thereby lengthening the relaxation ***,our observations indicate that in InAs/AlAs superlattices,the relaxation time increases as the excitation power increases,owing to the phonon bottleneck *** insights into InAs/AlAs superlattice carrier dynamics highlight their applicability in enhancing avalanche photodetectors,and may contribute to the optimized design of superlattices for specific applications.
Artificial Intelligence (AI) is transforming numerous domains, including bioinformatics and information extraction systems, by advancing data processing capabilities, enhancing precision, and facilitating automation. ...
详细信息
Although traditional research methods for intrusion detection can effectively prevent and mitigate issues such as data leaks to avoid severe consequences, existing intrusion detection technologies encounter limitation...
详细信息
The agriculture industry's production and food quality have been impacted by plant leaf diseases in recent years. Hence, it is vital to have a system that can automatically identify and diagnose diseases at an ini...
详细信息
1 Introduction As an emerging machine learning paradigm,unsupervised domain adaptation(UDA)aims to train an effective model for unlabeled target domain by leveraging knowledge from related but distribution-inconsisten...
详细信息
1 Introduction As an emerging machine learning paradigm,unsupervised domain adaptation(UDA)aims to train an effective model for unlabeled target domain by leveraging knowledge from related but distribution-inconsistent source *** of the existing UDA methods[2]align class-wise distributions resorting to target domain pseudo-labels,for which hard labels may be misguided by misclassifications while soft labels are confusing with trivial noises so that both of them tend to cause frustrating *** overcome such drawbacks,as shown in Fig.1,we propose to achieve UDA by performing self-adaptive label filtering learning(SALFL)from both the statistical and the geometrical perspectives,which filters out the misclassified pseudo-labels to reduce negative ***,the proposed SALFL firstly predicts labels for the target domain instances by graph-based random walking and then filters out those noise labels by self-adaptive learning strategy.
Gradient compression is a promising approach to alleviating the communication bottleneck in data parallel deep neural network (DNN) training by significantly reducing the data volume of gradients for synchronization. ...
详细信息
Gradient compression is a promising approach to alleviating the communication bottleneck in data parallel deep neural network (DNN) training by significantly reducing the data volume of gradients for synchronization. While gradient compression is being actively adopted by the industry (e.g., Facebook and AWS), our study reveals that there are two critical but often overlooked challenges: 1) inefficient coordination between compression and communication during gradient synchronization incurs substantial overheads, and 2) developing, optimizing, and integrating gradient compression algorithms into DNN systems imposes heavy burdens on DNN practitioners, and ad-hoc compression implementations often yield surprisingly poor system performance. In this paper, we propose a compression-aware gradient synchronization architecture, CaSync, which relies on flexible composition of basic computing and communication primitives. It is general and compatible with any gradient compression algorithms and gradient synchronization strategies and enables high-performance computation-communication pipelining. We further introduce a gradient compression toolkit, CompLL, to enable efficient development and automated integration of on-GPU compression algorithms into DNN systems with little programming burden. Lastly, we build a compression-aware DNN training framework HiPress with CaSync and CompLL. HiPress is open-sourced and runs on mainstream DNN systems such as MXNet, TensorFlow, and PyTorch. Evaluation via a 16-node cluster with 128 NVIDIA V100 GPUs and a 100 Gbps network shows that HiPress improves the training speed over current compression-enabled systems (e.g., BytePS-onebit, Ring-DGC and PyTorch-PowerSGD) by 9.8%-69.5% across six popular DNN models. IEEE
Based on the analysis of factors influencing the combat mission process, considering heterogeneous nodes, constraint relationships, and various influencing factors, this study proposes a command and control structure ...
详细信息
暂无评论