scholarly journals Compare the performance of the models in art classification

PLoS ONE ◽  
2021 ◽  
Vol 16 (3) ◽  
pp. e0248414
Author(s):  
Wentao Zhao ◽  
Dalin Zhou ◽  
Xinguo Qiu ◽  
Wei Jiang

Because large numbers of artworks are preserved in museums and galleries, much work must be done to classify these works into genres, styles and artists. Recent technological advancements have enabled an increasing number of artworks to be digitized. Thus, it is necessary to teach computers to analyze (e.g., classify and annotate) art to assist people in performing such tasks. In this study, we tested 7 different models on 3 different datasets under the same experimental setup to compare their art classification performances when either using or not using transfer learning. The models were compared based on their abilities for classifying genres, styles and artists. Comparing the result with previous work shows that the model performance can be effectively improved by optimizing the model structure, and our results achieve state-of-the-art performance in all classification tasks with three datasets. In addition, we visualized the process of style and genre classification to help us understand the difficulties that computers have when tasked with classifying art. Finally, we used the trained models described above to perform similarity searches and obtained performance improvements.

2020 ◽  
Vol 6 (11) ◽  
pp. 127
Author(s):  
Ibrahem Kandel ◽  
Mauro Castelli ◽  
Aleš Popovič

The classification of the musculoskeletal images can be very challenging, mostly when it is being done in the emergency room, where a decision must be made rapidly. The computer vision domain has gained increasing attention in recent years, due to its achievements in image classification. The convolutional neural network (CNN) is one of the latest computer vision algorithms that achieved state-of-the-art results. A CNN requires an enormous number of images to be adequately trained, and these are always scarce in the medical field. Transfer learning is a technique that is being used to train the CNN by using fewer images. In this paper, we study the appropriate method to classify musculoskeletal images by transfer learning and by training from scratch. We applied six state-of-the-art architectures and compared their performance with transfer learning and with a network trained from scratch. From our results, transfer learning did increase the model performance significantly, and, additionally, it made the model less prone to overfitting.


2019 ◽  
Author(s):  
Henrique Freitas ◽  
Celso Luiz Mendes

The Roofline model gives insights about the performance behavior of applications bounded by either memory or processor limits, providing useful guidelines for performance improvements. This work uses the Roofline model on the analysis of the MGB model that simulates hydrological processes in largescale watersheds. Real-world input data are used to characterize the performance on two multicore architectures, one with only CPUs and one with CPUs/GPU. The MGB model performance is improved with optimizations for better memory use, and also with shared-memory (OpenMP) and GPU (OpenACC) parallelism. CPU performance achieves 42.51 % and 50.17 % of each system’s peak, whereas GPU performance is low due to overheads caused by the MGB model structure.


Author(s):  
Michael Withnall ◽  
Edvard Lindelöf ◽  
Ola Engkvist ◽  
Hongming Chen

We introduce Attention and Edge Memory schemes to the existing Message Passing Neural Network framework for graph convolution, and benchmark our approaches against eight different physical-chemical and bioactivity datasets from the literature. We remove the need to introduce <i>a priori</i> knowledge of the task and chemical descriptor calculation by using only fundamental graph-derived properties. Our results consistently perform on-par with other state-of-the-art machine learning approaches, and set a new standard on sparse multi-task virtual screening targets. We also investigate model performance as a function of dataset preprocessing, and make some suggestions regarding hyperparameter selection.


2020 ◽  
Author(s):  
Pathikkumar Patel ◽  
Bhargav Lad ◽  
Jinan Fiaidhi

During the last few years, RNN models have been extensively used and they have proven to be better for sequence and text data. RNNs have achieved state-of-the-art performance levels in several applications such as text classification, sequence to sequence modelling and time series forecasting. In this article we will review different Machine Learning and Deep Learning based approaches for text data and look at the results obtained from these methods. This work also explores the use of transfer learning in NLP and how it affects the performance of models on a specific application of sentiment analysis.


1999 ◽  
Vol 39 (9) ◽  
pp. 1-8 ◽  
Author(s):  
P. Harremoës ◽  
H. Madsen

Where is the balance between simplicity and complexity in model prediction of urban drainage structures? The calibration/verification approach to testing of model performance gives an exaggerated sense of certainty. Frequently, the model structure and the parameters are not identifiable by calibration/verification on the basis of the data series available, which generates elements of sheer guessing - unless the universality of the model is be based on induction, i.e. experience from the sum of all previous investigations. There is a need to deal more explicitly with uncertainty and to incorporate that in the design, operation and control of urban drainage structures.


2008 ◽  
Vol 600-603 ◽  
pp. 895-900 ◽  
Author(s):  
Anant K. Agarwal ◽  
Albert A. Burk ◽  
Robert Callanan ◽  
Craig Capell ◽  
Mrinal K. Das ◽  
...  

In this paper, we review the state of the art of SiC switches and the technical issues which remain. Specifically, we will review the progress and remaining challenges associated with SiC power MOSFETs and BJTs. The most difficult issue when fabricating MOSFETs has been an excessive variation in threshold voltage from batch to batch. This difficulty arises due to the fact that the threshold voltage is determined by the difference between two large numbers, namely, a large fixed oxide charge and a large negative charge in the interface traps. There may also be some significant charge captured in the bulk traps in SiC and SiO2. The effect of recombination-induced stacking faults (SFs) on majority carrier mobility has been confirmed with 10 kV Merged PN Schottky (MPS) diodes and MOSFETs. The same SFs have been found to be responsible for degradation of BJTs.


2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Young-Gon Kim ◽  
Sungchul Kim ◽  
Cristina Eunbee Cho ◽  
In Hye Song ◽  
Hee Jin Lee ◽  
...  

AbstractFast and accurate confirmation of metastasis on the frozen tissue section of intraoperative sentinel lymph node biopsy is an essential tool for critical surgical decisions. However, accurate diagnosis by pathologists is difficult within the time limitations. Training a robust and accurate deep learning model is also difficult owing to the limited number of frozen datasets with high quality labels. To overcome these issues, we validated the effectiveness of transfer learning from CAMELYON16 to improve performance of the convolutional neural network (CNN)-based classification model on our frozen dataset (N = 297) from Asan Medical Center (AMC). Among the 297 whole slide images (WSIs), 157 and 40 WSIs were used to train deep learning models with different dataset ratios at 2, 4, 8, 20, 40, and 100%. The remaining, i.e., 100 WSIs, were used to validate model performance in terms of patch- and slide-level classification. An additional 228 WSIs from Seoul National University Bundang Hospital (SNUBH) were used as an external validation. Three initial weights, i.e., scratch-based (random initialization), ImageNet-based, and CAMELYON16-based models were used to validate their effectiveness in external validation. In the patch-level classification results on the AMC dataset, CAMELYON16-based models trained with a small dataset (up to 40%, i.e., 62 WSIs) showed a significantly higher area under the curve (AUC) of 0.929 than those of the scratch- and ImageNet-based models at 0.897 and 0.919, respectively, while CAMELYON16-based and ImageNet-based models trained with 100% of the training dataset showed comparable AUCs at 0.944 and 0.943, respectively. For the external validation, CAMELYON16-based models showed higher AUCs than those of the scratch- and ImageNet-based models. Model performance for slide feasibility of the transfer learning to enhance model performance was validated in the case of frozen section datasets with limited numbers.


Electronics ◽  
2021 ◽  
Vol 10 (15) ◽  
pp. 1807
Author(s):  
Sascha Grollmisch ◽  
Estefanía Cano

Including unlabeled data in the training process of neural networks using Semi-Supervised Learning (SSL) has shown impressive results in the image domain, where state-of-the-art results were obtained with only a fraction of the labeled data. The commonality between recent SSL methods is that they strongly rely on the augmentation of unannotated data. This is vastly unexplored for audio data. In this work, SSL using the state-of-the-art FixMatch approach is evaluated on three audio classification tasks, including music, industrial sounds, and acoustic scenes. The performance of FixMatch is compared to Convolutional Neural Networks (CNN) trained from scratch, Transfer Learning, and SSL using the Mean Teacher approach. Additionally, a simple yet effective approach for selecting suitable augmentation methods for FixMatch is introduced. FixMatch with the proposed modifications always outperformed Mean Teacher and the CNNs trained from scratch. For the industrial sounds and music datasets, the CNN baseline performance using the full dataset was reached with less than 5% of the initial training data, demonstrating the potential of recent SSL methods for audio data. Transfer Learning outperformed FixMatch only for the most challenging dataset from acoustic scene classification, showing that there is still room for improvement.


2021 ◽  
Vol 9 (5) ◽  
pp. 467
Author(s):  
Mostafa Farrag ◽  
Gerald Corzo Perez ◽  
Dimitri Solomatine

Many grid-based spatial hydrological models suffer from the complexity of setting up a coherent spatial structure to calibrate such a complex, highly parameterized system. There are essential aspects of model-building to be taken into account: spatial resolution, the routing equation limitations, and calibration of spatial parameters, and their influence on modeling results, all are decisions that are often made without adequate analysis. In this research, an experimental analysis of grid discretization level, an analysis of processes integration, and the routing concepts are analyzed. The HBV-96 model is set up for each cell, and later on, cells are integrated into an interlinked modeling system (Hapi). The Jiboa River Basin in El Salvador is used as a case study. The first concept tested is the model structure temporal responses, which are highly linked to the runoff dynamics. By changing the runoff generation model description, we explore the responses to events. Two routing models are considered: Muskingum, which routes the runoff from each cell following the river network, and Maxbas, which routes the runoff directly to the outlet. The second concept is the spatial representation, where the model is built and tested for different spatial resolutions (500 m, 1 km, 2 km, and 4 km). The results show that the spatial sensitivity of the resolution is highly linked to the routing method, and it was found that routing sensitivity influenced the model performance more than the spatial discretization, and allowing for coarser discretization makes the model simpler and computationally faster. Slight performance improvement is gained by using different parameters’ values for each cell. It was found that the 2 km cell size corresponds to the least model error values. The proposed hydrological modeling codes have been published as open-source.


2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Young Jae Kim ◽  
Jang Pyo Bae ◽  
Jun-Won Chung ◽  
Dong Kyun Park ◽  
Kwang Gi Kim ◽  
...  

AbstractWhile colorectal cancer is known to occur in the gastrointestinal tract. It is the third most common form of cancer of 27 major types of cancer in South Korea and worldwide. Colorectal polyps are known to increase the potential of developing colorectal cancer. Detected polyps need to be resected to reduce the risk of developing cancer. This research improved the performance of polyp classification through the fine-tuning of Network-in-Network (NIN) after applying a pre-trained model of the ImageNet database. Random shuffling is performed 20 times on 1000 colonoscopy images. Each set of data are divided into 800 images of training data and 200 images of test data. An accuracy evaluation is performed on 200 images of test data in 20 experiments. Three compared methods were constructed from AlexNet by transferring the weights trained by three different state-of-the-art databases. A normal AlexNet based method without transfer learning was also compared. The accuracy of the proposed method was higher in statistical significance than the accuracy of four other state-of-the-art methods, and showed an 18.9% improvement over the normal AlexNet based method. The area under the curve was approximately 0.930 ± 0.020, and the recall rate was 0.929 ± 0.029. An automatic algorithm can assist endoscopists in identifying polyps that are adenomatous by considering a high recall rate and accuracy. This system can enable the timely resection of polyps at an early stage.


Sign in / Sign up

Export Citation Format

Share Document