Classification of Small Sets of Images with Pre-trained Neural Networks

Petrovska, Biserka; Stojanović, Igor; Atanasova-Pacemska, Tatjana

doi:10.5815/ijem.2018.04.05

Cited by 6 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…NNs are a different series of models than ML, and there are many reasons for this, the most famous of which is the cost of running the algorithms on the machine .The pre-trained model may not be very accurate, but it saves a lot of effort to produce a good model [32][33]. It can be used for extracted features.…”

Section: Pre-trained Models In Deep Learningmentioning

confidence: 99%

Review Study on Artificial Intelligence Tools for Attention Deficit Hyperactivity Disorder Identification

2023

International Journal of Telecommunications

View full text Add to dashboard Cite

Attention deficit hyperactivity disorder (ADHD) is the most common developmental disease in childhood and adolescence. Most similar previous ADHD classification studies have only classified ADHD and normal categories. This study was the first attempt to classify adults with ADHD by means of a support vector machine indicating that classification by nonlinear methods is possible in the context of clinical populations. Some of the guidelines were used in this systematic review to analyze the studies most relevant to the diagnosis of ADHD using the Deep learning (DL) approach. This paper also attempted to look at the process of ADHD development; what are the associated problems? And how many other children and adults are affected by such problems around the world the basis for understanding ADHD more accurately in order to develop a better multi-modal medical and/or non-medical intervention plan and some classification technique of ADHD will be presented. The diagnosis of it includes conducting some psychological tests, but it may not give accurate results, so the diagnosis is made using non-invasive imaging, which is called Magnetic Resonance Imaging (MRI). For this purpose and according to the medical imaging system, first, the images are pre-processed, and then 3D images are taken and this is done using the ADHD-200 training data set. This shows the best results and classification with high accuracy.

show abstract

Section: Pre-trained Models In Deep Learningmentioning

confidence: 99%

Review Study on Artificial Intelligence Tools for Attention Deficit Hyperactivity Disorder Identification

2023

International Journal of Telecommunications

View full text Add to dashboard Cite

show abstract

“…Machine learning tends to perform poorly when the amount of training data is rare [8,9]. The fundamental problem of few-sample learning is that the amount of target task data is not enough, and it is difficult to train a robust learning model.…”

Section: Few-shot Learningmentioning

confidence: 99%

FeatureGAN: Combining GAN and Autoencoder for Pavement Crack Image Data Augmentations

Zhang¹,

Peng²,

Al‐Huda³

et al. 2022

IJIGSP

View full text Add to dashboard Cite

In the pavement crack segmentation task, the accurate pixel-level labeling for the fully supervised training of deep neural networks Deep Neural Networks (DNN) is challenging. Although cracks often exhibit low-level image characters in terms of edges, there might be various high-level background information based on the complex pavement conditions. In practice, crack samples containing various semantic backgrounds are scarce. To overcome these problems, we propose a novel method for augmenting the training data for DNN based crack segmentation task. It employs the generative adversarial network Generative Adversarial Networks (GAN), which utilizes a crack-free image, a crack image, and a corresponding image mask to generate a new crack image. In combination with an auto-encoder, the proposed GAN can be used to train crack segmentation networks. By creating a manual mask, no additional crack images are required to be labeled, and data augmentation and annotation are achieved simultaneously. Our experiments are conducted on two public datasets using five segmentation models to verify the effectiveness of the proposed method. Experimental results demonstrate that the proposed method is effective for crack segmentation.

show abstract

“…The difference between traditional machine learning and transfer learning is found in Figure 1. Nowadays, transfer learning has been applied to robotics [2,3] image classification [4,5], sentiment classification [6], game technology [7,8] and text classification [9]. Generally, the type of transfer learning used in deep learning is a pre-trained network.…”

Section: Introductionmentioning

confidence: 99%

Transfer learning with multiple pre-trained network for fundus classification

2020

View full text Add to dashboard Cite

Transfer learning (TL) is a technique of reuse and modify a pre-trained network. It reuses feature extraction layer at a pre-trained network. A target domain in TL obtains the features knowledge from the source domain. TL modified classification layer at a pre-trained network. The target domain can do new tasks according to a purpose. In this article, the target domain is fundus image classification includes normal and neovascularization. Data consist of 100 patches. The comparison of training and validation data was 70:30. The selection of training and validation data is done randomly. Steps of TL i.e load pre-trained networks, replace final layers, train the network, and assess network accuracy. First, the pre-trained network is a layer configuration of the convolutional neural network architecture. Pre-trained network used are AlexNet, VGG16, VGG19, ResNet50, ResNet101, GoogLeNet, Inception-V3, InceptionResNetV2, and squeezenet. Second, replace the final layer is to replace the last three layers. They are fully connected layer, softmax, and output layer. The layer is replaced with a fully connected layer that classifies according to number of classes. Furthermore, it's followed by a softmax and output layer that matches with the target domain. Third, we trained the network. Networks were trained to produce optimal accuracy. In this section, we use gradient descent algorithm optimization. Fourth, assess network accuracy. The experiment results show a testing accuracy between 80% and 100%.

show abstract

Classification of Small Sets of Images with Pre-trained Neural Networks

Cited by 6 publications

References 19 publications

Review Study on Artificial Intelligence Tools for Attention Deficit Hyperactivity Disorder Identification

Review Study on Artificial Intelligence Tools for Attention Deficit Hyperactivity Disorder Identification

FeatureGAN: Combining GAN and Autoencoder for Pavement Crack Image Data Augmentations

Transfer learning with multiple pre-trained network for fundus classification

Contact Info

Product

Resources

About