Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userSelective Synthetic Augmentation with Quality Assurance
71
0
0.0
(
0
)
Added by
Yuan Xue
Publication date
2019
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Supervised training of an automated medical image analysis system often requires a large amount of expert annotations that are hard to collect. Moreover, the proportions of data available across different classes may be highly imbalanced for rare diseases. To mitigate these issues, we investigate a novel data augmentation pipeline that selectively adds new synthetic images generated by conditional Adversarial Networks (cGANs), rather than extending directly the training set with synthetic images. The selection mechanisms that we introduce to the synthetic augmentation pipeline are motivated by the observation that, although cGAN-generated images can be visually appealing, they are not guaranteed to contain essential features for classification performance improvement. By selecting synthetic images based on the confidence of their assigned labels and their feature similarity to real labeled images, our framework provides quality assurance to synthetic augmentation by ensuring that adding the selected synthetic images to the training set will improve performance. We evaluate our model on a medical histopathology dataset, and two natural image classification benchmarks, CIFAR10 and SVHN. Results on these datasets show significant and consistent improvements in classification performance (with 6.8%, 3.9%, 1.6% higher accuracy, respectively) by leveraging cGAN generated images with selective augmentation.
rate research
Read More
Contemporary Artificial Intelligence technologies allow for the employment of Computer Vision to discern good crops from bad, providing a step in the pipeline of selecting healthy fruit from undesirable fruit, such as those which are mouldy or gangrenous. State-of-the-art works in the field report high accuracy results on small datasets (<1000 images), which are not representative of the population regarding real-world usage. The goals of this study are to further enable real-world usage by improving generalisation with data augmentation as well as to reduce overfitting and energy usage through model pruning. In this work, we suggest a machine learning pipeline that combines the ideas of fine-tuning, transfer learning, and generative model-based training data augmentation towards improving fruit quality image classification. A linear network topology search is performed to tune a VGG16 lemon quality classification model using a publicly-available dataset of 2690 images. We find that appending a 4096 neuron fully connected layer to the convolutional layers leads to an image classification accuracy of 83.77%. We then train a Conditional Generative Adversarial Network on the training data for 2000 epochs, and it learns to generate relatively realistic images. Grad-CAM analysis of the model trained on real photographs shows that the synthetic images can exhibit classifiable characteristics such as shape, mould, and gangrene. A higher image classification accuracy of 88.75% is then attained by augmenting the training with synthetic images, arguing that Conditional Generative Adversarial Networks have the ability to produce new data to alleviate issues of data scarcity. Finally, model pruning is performed via polynomial decay, where we find that the Conditional GAN-augmented classification network can retain 81.16% classification accuracy when compressed to 50% of its original size.
An analysis software was developed for the high aspect ratio optical scanning system in the Detec- tor Laboratory of the University of Helsinki and the Helsinki Institute of Physics. The system is used e.g. in the quality assurance of the GEM-TPC detectors being developed for the beam diagnostics system of the SuperFRS at future FAIR facility. The software was tested by analyzing five CERN standard GEM foils scanned with the optical scanning system. The measurement uncertainty of the diameter of the GEM holes and the pitch of the hole pattern was found to be 0.5 {mu}m and 0.3 {mu}m, respectively. The software design and the performance are discussed. The correlation between the GEM hole size distribution and the corresponding gain variation was studied by comparing them against a detailed gain mapping of a foil and a set of six lower precision control measurements. It can be seen that a qualitative estimation of the behavior of the local variation in gain across the GEM foil can be made based on the measured sizes of the outer and inner holes.
The MAJORANA DEMONSTRATOR is an experiment constructed to search for neutrinoless double-beta decays in germanium-76 and to demonstrate the feasibility to deploy a large-scale experiment in a phased and modular fashion. It consists of two modular arrays of natural and $^{76}$Ge-enriched germanium detectors totalling 44.1 kg, located at the 4850 level of the Sanford Underground Research Facility in Lead, South Dakota, USA. Any neutrinoless double-beta decay search requires a thorough understanding of the background and the signal energy spectra. The various techniques employed to ensure the integrity of the measured spectra are discussed. Data collection is monitored with a thorough set of checks, and subsequent careful analysis is performed to qualify the data for higher level physics analysis. Instrumental background events are tagged for removal, and problematic channels are removed from consideration as necessary.
The Atacama Large mm and sub-mm Array (ALMA) radio observatory is one of the worlds largest astronomical projects. After the very successful conclusion of the first observation cycles Early Science Cycles 0 and 1, the ALMA project can report many successes and lessons learned. The science data taken interleaved with commissioning tests for the still continuing addition of new capabilities has already resulted in numerous publications in high-profile journals. The increasing data volume and complexity are challenging but under control. The radio-astronomical data analysis package Common Astronomy Software Applications (CASA) has played a crucial role in this effort. This article describes the implementation of the ALMA data quality assurance system, in particular the level 2 which is based on CASA, and the lessons learned.
The main results of the quality assurance tests performed on the Resistive Plate Chamber used by the ATLAS experiment at LHC as muon trigger chambers are reported and discussed. Since July 2004, about 270 RPC units has been certified at INFN Lecce site and delivered to CERN, for being integrated in the final muon station of the ATLAS barrel region. We show the key RPC characteristics which qualify the performance of this detector technology as muon trigger chamber in the harsh LHC enviroments. These are dark current, chamber efficiency, noise rate, gas volume tomography, and gas leakage.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
