Advancing diagnostic performance and clinical usability of neural networks via adversarial training and dual batch normalization.
Ontology highlight
ABSTRACT:
SUBMITTER: Han T
PROVIDER: S-EPMC8280105 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Spiking Neural Networks (SNNs) have recently emerged as an alternative to deep learning owing to sparse, asynchronous and binary event (or spike) driven processing, that can yield huge energy efficiency benefits on neuromorphic hardware. However, SNNs convey temporally-varying spike activation through time that is likely to induce a large variation of forward activation and backward gradients, resulting in unstable training. To address this training issue in SNNs, we revisit Batch Normalization (BN) and propose a temporal Batch Normalization Through Time (BNTT) technique. Different from previous BN techniques with SNNs, we find that varying the BN parameters at every time-step allows the model to learn the time-varying input distribution better. Specifically, our proposed BNTT decouples the parameters in a BNTT layer along the time axis to capture the temporal dynamics of spikes. We demonstrate BNTT on CIFAR-10, CIFAR-100, Tiny-ImageNet, event-driven DVS-CIFAR10 datasets, and Sequential MNIST and show near state-of-the-art performance. We conduct comprehensive analysis on the temporal characteristic of BNTT and showcase interesting benefits toward robustness against random and adversarial noise. Further, by monitoring the learnt parameters of BNTT, we find that we can do temporal early exit. That is, we can reduce the inference latency by ~5 - 20 time-steps from the original training latency. The code has been released at https://github.com/Intelligent-Computing-Lab-Yale/BNTT-Batch-Normalization-Through-Time.
| S-EPMC8695433 | biostudies-literature
Project description:MotivationUntargeted metabolomics by mass spectrometry is the method of choice for unbiased analysis of molecules in complex samples of biological, clinical or environmental relevance. The exceptional versatility and sensitivity of modern high-resolution instruments allows profiling of thousands of known and unknown molecules in parallel. Inter-batch differences constitute a common and unresolved problem in untargeted metabolomics, and hinder the analysis of multi-batch studies or the intercomparison of experiments.ResultsWe present a new method, Regularized Adversarial Learning Preserving Similarity (RALPS), for the normalization of multi-batch untargeted metabolomics data. RALPS builds on deep adversarial learning with a three-term loss function that mitigates batch effects while preserving biological identity, spectral properties and coefficients of variation. Using two large metabolomics datasets, we showcase the superior performance of RALPS as compared with six state-of-the-art methods for batch correction. Further, we demonstrate that RALPS scales well, is robust, deals with missing values and can handle different experimental designs.Availability and implementationhttps://github.com/zamboni-lab/RALPS.Supplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC9978579 | biostudies-literature
Project description:Annotating cell identities is a common bottleneck in the analysis of single-cell genomics experiments. Here, we present scNym, a semisupervised, adversarial neural network that learns to transfer cell identity annotations from one experiment to another. scNym takes advantage of information in both labeled data sets and new, unlabeled data sets to learn rich representations of cell identity that enable effective annotation transfer. We show that scNym effectively transfers annotations across experiments despite biological and technical differences, achieving performance superior to existing methods. We also show that scNym models can synthesize information from multiple training and target data sets to improve performance. We show that in addition to high accuracy, scNym models are well calibrated and interpretable with saliency methods.
| S-EPMC8494222 | biostudies-literature
Project description:Neural networks based on nanodevices, such as metal oxide memristors, phase change memories, and flash memory cells, have generated considerable interest for their increased energy efficiency and density in comparison to graphics processing units (GPUs) and central processing units (CPUs). Though immense acceleration of the training process can be achieved by leveraging the fact that the time complexity of training does not scale with the network size, it is limited by the space complexity of stochastic gradient descent, which grows quadratically. The main objective of this work is to reduce this space complexity by using low-rank approximations of stochastic gradient descent. This low spatial complexity combined with streaming methods allows for significant reductions in memory and compute overhead, opening the door for improvements in area, time and energy efficiency of training. We refer to this algorithm and architecture to implement it as the streaming batch eigenupdate (SBE) approach.
| S-EPMC6691093 | biostudies-literature
Project description:BackgroundDeep neural networks (DNNs) are widely investigated in medical image classification to achieve automated support for clinical diagnosis. It is necessary to evaluate the robustness of medical DNN tasks against adversarial attacks, as high-stake decision-making will be made based on the diagnosis. Several previous studies have considered simple adversarial attacks. However, the vulnerability of DNNs to more realistic and higher risk attacks, such as universal adversarial perturbation (UAP), which is a single perturbation that can induce DNN failure in most classification tasks has not been evaluated yet.MethodsWe focus on three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) and investigate their vulnerability to the seven model architectures of UAPs.ResultsWe demonstrate that DNNs are vulnerable to both nontargeted UAPs, which cause a task failure resulting in an input being assigned an incorrect class, and to targeted UAPs, which cause the DNN to classify an input into a specific class. The almost imperceptible UAPs achieved?>?80% success rates for nontargeted and targeted attacks. The vulnerability to UAPs depended very little on the model architecture. Moreover, we discovered that adversarial retraining, which is known to be an effective method for adversarial defenses, increased DNNs' robustness against UAPs in only very few cases.ConclusionUnlike previous assumptions, the results indicate that DNN-based clinical diagnosis is easier to deceive because of adversarial attacks. Adversaries can cause failed diagnoses at lower costs (e.g., without consideration of data distribution); moreover, they can affect the diagnosis. The effects of adversarial defenses may not be limited. Our findings emphasize that more careful consideration is required in developing DNNs for medical imaging and their practical applications.
| S-EPMC7792111 | biostudies-literature
Project description:Neural networks enjoy widespread success in both research and industry and, with the advent of quantum technology, it is a crucial challenge to design quantum neural networks for fully quantum learning tasks. Here we propose a truly quantum analogue of classical neurons, which form quantum feedforward neural networks capable of universal quantum computation. We describe the efficient training of these networks using the fidelity as a cost function, providing both classical and efficient quantum implementations. Our method allows for fast optimisation with reduced memory requirements: the number of qudits required scales with only the width, allowing deep-network optimisation. We benchmark our proposal for the quantum task of learning an unknown unitary and find remarkable generalisation behaviour and a striking robustness to noisy training data.
| S-EPMC7010779 | biostudies-literature
Project description:The reason for the existence of adversarial samples is still barely understood. Here, we explore the transferability of learned features to Out-of-Distribution (OoD) classes. We do this by assessing neural networks' capability to encode the existing features, revealing an intriguing connection with adversarial attacks and defences. The principal idea is that, "if an algorithm learns rich features, such features should represent Out-of-Distribution classes as a combination of previously learned In-Distribution (ID) classes". This is because OoD classes usually share several regular features with ID classes, given that the features learned are general enough. We further introduce two metrics to assess the transferred features representing OoD classes. One is based on inter-cluster validation techniques, while the other captures the influence of a class over learned features. Experiments suggest that several adversarial defences decrease the attack accuracy of some attacks and improve the transferability-of-features as measured by our metrics. Experiments also reveal a relationship between the proposed metrics and adversarial attacks (a high Pearson correlation coefficient and low p-value). Further, statistical tests suggest that several adversarial defences, in general, significantly improve transferability. Our tests suggests that models having a higher transferability-of-features have generally higher robustness against adversarial attacks. Thus, the experiments suggest that the objectives of adversarial machine learning might be much closer to domain transfer learning, as previously thought.
| S-EPMC9045664 | biostudies-literature
Project description:Since the variables inherent to various diseases cannot be controlled directly in humans, behavioral dysfunctions have been examined in model organisms, leading to better understanding their underlying mechanisms. However, because the spatial and temporal scales of animal locomotion vary widely among species, conventional statistical analyses cannot be used to discover knowledge from the locomotion data. We propose a procedure to automatically discover locomotion features shared among animal species by means of domain-adversarial deep neural networks. Our neural network is equipped with a function which explains the meaning of segments of locomotion where the cross-species features are hidden by incorporating an attention mechanism into the neural network, regarded as a black box. It enables us to formulate a human-interpretable rule about the cross-species locomotion feature and validate it using statistical tests. We demonstrate the versatility of this procedure by identifying locomotion features shared across different species with dopamine deficiency, namely humans, mice, and worms, despite their evolutionary differences.
| S-EPMC8448872 | biostudies-literature
Project description:Digital breast tomosynthesis (DBT) is a quasi-three-dimensional imaging modality that can reduce false negatives and false positives in mass lesion detection caused by overlapping breast tissue in conventional two-dimensional (2D) mammography. The patient dose of a DBT scan is similar to that of a single 2D mammogram, while acquisition of each projection view adds detector readout noise. The noise is propagated to the reconstructed DBT volume, possibly obscuring subtle signs of breast cancer such as microcalcifications (MCs). This study developed a deep convolutional neural network (DCNN) framework for denoising DBT images with a focus on improving the conspicuity of MCs as well as preserving the ill-defined margins of spiculated masses and normal tissue textures. We trained the DCNN using a weighted combination of mean squared error (MSE) loss and adversarial loss. We configured a dedicated x-ray imaging simulator in combination with digital breast phantoms to generate realistic in silico DBT data for training. We compared the DCNN training between using digital phantoms and using real physical phantoms. The proposed denoising method improved the contrast-to-noise ratio (CNR) and detectability index (d') of the simulated MCs in the validation phantom DBTs. These performance measures improved with increasing training target dose and training sample size. Promising denoising results were observed on the transferability of the digital-phantom-trained denoiser to DBT reconstructed with different techniques and on a small independent test set of human subject DBT images.
| S-EPMC8274391 | biostudies-literature
Project description:Big genomic data and artificial intelligence (AI) are ushering in an era of precision medicine, providing opportunities to study previously under-represented subtypes and rare diseases rather than categorize them as variances. However, clinical researchers face challenges in accessing such novel technologies as well as reliable methods to study small datasets or subcohorts with unique phenotypes. To address this need, we developed an integrative approach, GAiN, to capture patterns of gene expression from small datasets on the basis of an ensemble of generative adversarial networks (GANs) while leveraging big population data. Where conventional biostatistical methods fail, GAiN reliably discovers differentially expressed genes (DEGs) and enriched pathways between two cohorts with limited numbers of samples (n = 10) when benchmarked against a gold standard. GAiN is freely available at GitHub. Thus, GAiN may serve as a crucial tool for gene expression analysis in scenarios with limited samples, as in the context of rare diseases, under-represented populations, or limited investigator resources.
| S-EPMC10873154 | biostudies-literature