The detection of new multiple sclerosis (MS) lesions is an important marker of the evolution of the disease. The applicability of learning-based methods could automate this task efficiently. However, the lack of annotated longitudinal data with new-appearing lesions is a limiting factor for the training of robust and generalizing models. In this work, we describe a deep-learning-based pipeline addressing the challenging task of detecting and segmenting new MS lesions. First, we propose to use transfer-learning from a model trained on a segmentation task using single time-points. Therefore, we exploit knowledge from an easier task and for which more annotated datasets are available. Second, we propose a data synthesis strategy to generate realistic longitudinal time-points with new lesions using single time-point scans. In this way, we pretrain our detection model on large synthetic annotated datasets. Finally, we use a data-augmentation technique designed to simulate data diversity in MRI. By doing that, we increase the size of the available small annotated longitudinal datasets. Our ablation study showed that each contribution lead to an enhancement of the segmentation accuracy. Using the proposed pipeline, we obtained the best score for the segmentation and the detection of new MS lesions in the MSSEG2 MICCAI challenge.
Estimated uncertainty by approximate posteriors in Bayesian neural networks are prone to miscalibration, which leads to overconfident predictions in critical tasks that have a clear asymmetric cost or significant losses. Here, we extend the approximate inference for the loss-calibrated Bayesian framework to dropweights based Bayesian neural networks by maximising expected utility over a model posterior to calibrate uncertainty in deep learning. Furthermore, we show that decisions informed by loss-calibrated uncertainty can improve diagnostic performance to a greater extent than straightforward alternatives. We propose Maximum Uncertainty Calibration Error (MUCE) as a metric to measure calibrated confidence, in addition to its prediction especially for high-risk applications, where the goal is to minimise the worst-case deviation between error and estimated uncertainty. In experiments, we show the correlation between error in prediction and estimated uncertainty by interpreting Wasserstein distance as the accuracy of prediction. We evaluated the effectiveness of our approach to detecting Covid-19 from X-Ray images. Experimental results show that our method reduces miscalibration considerably, without impacting the models accuracy and improves reliability of computer-based diagnostics.
Automatic evaluation of essay (AES) and also called automatic essay scoring has become a severe problem due to the rise of online learning and evaluation platforms such as Coursera, Udemy, Khan academy, and so on. Researchers have recently proposed many techniques for automatic evaluation. However, many of these techniques use hand-crafted features and thus are limited from the feature representation point of view. Deep learning has emerged as a new paradigm in machine learning which can exploit the vast data and identify the features useful for essay evaluation. To this end, we propose a novel architecture based on recurrent networks (RNN) and convolution neural network (CNN). In the proposed architecture, the multichannel convolutional layer learns and captures the contextual features of the word n-gram from the word embedding vectors and the essential semantic concepts to form the feature vector at essay level using max-pooling operation. A variant of RNN called Bi-gated recurrent unit (BGRU) is used to access both previous and subsequent contextual representations. The experiment was carried out on eight data sets available on Kaggle for the task of AES. The experimental results show that our proposed system achieves significantly higher grading accuracy than other deep learning-based AES systems and also other state-of-the-art AES systems.
In most popular public accessible cryptocurrency systems, the mining pool plays a key role because mining cryptocurrency with the mining pool turns the non-profitable situation into profitable for individual miners. In many recent novel blockchain consensuses, the deep learning training procedure becomes the task for miners to prove their workload, thus the computation power of miners will not purely be spent on the hash puzzle. In this way, the hardware and energy will support the blockchain service and deep learning training simultaneously. While the incentive of miners is to earn tokens, individual miners are motivated to join mining pools to become more competitive. In this paper, we are the first to demonstrate a mining pool solution for novel consensuses based on deep learning. The mining pool manager partitions the full searching space into subspaces and all miners are scheduled to collaborate on the Neural Architecture Search (NAS) tasks in the assigned subspace. Experiments demonstrate that the performance of this type of mining pool is more competitive than an individual miner. Due to the uncertainty of miners' behaviors, the mining pool manager checks the standard deviation of the performance of high reward miners and prepares backup miners to ensure the completion of the tasks of high reward miners.
This work is concerned with the following fundamental question in scientific machine learning: Can deep-learning-based methods solve noise-free inverse problems to near-perfect accuracy? Positive evidence is provided for the first time, focusing on a prototypical computed tomography (CT) setup. We demonstrate that an iterative end-to-end network scheme enables reconstructions close to numerical precision, comparable to classical compressed sensing strategies. Our results build on our winning submission to the recent AAPM DL-Sparse-View CT Challenge. Its goal was to identify the state-of-the-art in solving the sparse-view CT inverse problem with data-driven techniques. A specific difficulty of the challenge setup was that the precise forward model remained unknown to the participants. Therefore, a key feature of our approach was to initially estimate the unknown fanbeam geometry in a data-driven calibration step. Apart from an in-depth analysis of our methodology, we also demonstrate its state-of-the-art performance on the open-access real-world dataset LoDoPaB CT.
The belief function approach to uncertainty quantification as proposed in the Demspter-Shafer theory of evidence is established upon the general mathematical models for set-valued observations, called random sets. Set-valued predictions are the most natural representations of uncertainty in machine learning. In this paper, we introduce a concept called epistemic deep learning based on the random-set interpretation of belief functions to model epistemic learning in deep neural networks. We propose a novel random-set convolutional neural network for classification that produces scores for sets of classes by learning set-valued ground truth representations. We evaluate different formulations of entropy and distance measures for belief functions as viable loss functions for these random-set networks. We also discuss methods for evaluating the quality of epistemic predictions and the performance of epistemic random-set neural networks. We demonstrate through experiments that the epistemic approach produces better performance results when compared to traditional approaches of estimating uncertainty.
In this paper, we review physics- and data-driven reconstruction techniques for simultaneous positron emission tomography (PET) / magnetic resonance imaging (MRI) systems, which have significant advantages for clinical imaging of cancer, neurological disorders, and heart disease. These reconstruction approaches utilize priors, either structural or statistical, together with a physics-based description of the PET system response. However, due to the nested representation of the forward problem, direct PET/MRI reconstruction is a nonlinear problem. We elucidate how a multi-faceted approach accommodates hybrid data- and physics-driven machine learning for reconstruction of 3D PET/MRI, summarizing important deep learning developments made in the last 5 years to address attenuation correction, scattering, low photon counts, and data consistency. We also describe how applications of these multi-modality approaches extend beyond PET/MRI to improving accuracy in radiation therapy planning. We conclude by discussing opportunities for extending the current state-of-the-art following the latest trends in physics- and deep learning-based computational imaging and next-generation detector hardware.
While a multitude of studies have been conducted on graph drawing, many existing methods only focus on optimizing a single aesthetic aspect of graph layouts, which can lead to sub-optimal results. There are a few existing methods that have attempted to develop a flexible solution for optimizing different aesthetic aspects measured by different aesthetic criteria. Furthermore, thanks to the significant advance in deep learning techniques, several deep learning-based layout methods were proposed recently. These methods have demonstrated the advantages of deep learning approaches for graph drawing. However, none of these existing methods can be directly applied to optimizing non-differentiable criteria without special accommodation. In this work, we propose a novel Generative Adversarial Network (GAN) based deep learning framework for graph drawing, called SmartGD, which can optimize different quantitative aesthetic goals, regardless of their differentiability. To demonstrate the effectiveness and efficiency of SmartGD, we conducted experiments on minimizing stress, minimizing edge crossing, maximizing crossing angle, maximizing shape-based metrics, and a combination of multiple aesthetics. Compared with several popular graph drawing algorithms, the experimental results show that SmartGD achieves good performance both quantitatively and qualitatively.
Security of information passing through the Internet is threatened by today’s most advanced malware ranging from orchestrated botnets to simpler polymorphic worms. These threats, as examples of zero-day attacks, are able to change their behavior several times in the early phases of their existence to bypass the network intrusion detection systems (NIDS). In fact, even well-designed, and frequently-updated signaturebased NIDS cannot detect the zero-day treats due to the lack of an adequate signature database, adaptive to intelligent attacks on the Internet. More importantly, having an NIDS, it should be tested on malicious traffic dataset that not only represents known attacks, but also can to some extent reflect the characteristics of unknown, zero-day attacks. Generating such traffic is identified in the literature as one of the main obstacles for evaluating the effectiveness of NIDS. To address these issues, we introduce RNNIDS that applies Recurrent Neural Networks (RNNs) to find complex patterns in attacks and generate similar ones. In this regard, for the first time, we demonstrate that RNNs are helpful to generate new, unseen mutants of attacks as well as synthetic signatures from the most advanced malware to improve the intrusion detection rate. Besides, to further enhance the design of an NIDS, RNNs can be employed to generate malicious datasets containing, e.g., unseen mutants of a malware. To evaluate the feasibility of our approaches, we conduct extensive experiments by incorporating publicly available datasets, where we show a considerable improvement in the detection rate of an off-the-shelf NIDS (up to 16.67%).
In this report, we presents low-complexity deep learning frameworks for acoustic scene classification (ASC). The proposed frameworks can be separated into four main steps: Front-end spectrogram extraction, online data augmentation, back-end classification, and late fusion of predicted probabilities. In particular, we initially transform audio recordings into Mel, Gammatone, and CQT spectrograms. Next, data augmentation methods of Random Cropping, Specaugment, and Mixup are then applied to generate augmented spectrograms before being fed into deep learning based classifiers. Finally, to achieve the best performance, we fuse probabilities which obtained from three individual classifiers, which are independently-trained with three type of spectrograms. Our experiments conducted on DCASE 2022 Task 1 Development dataset have fullfiled the requirement of low-complexity and achieved the best classification accuracy of 60.1%, improving DCASE baseline by 17.2%.
The bone age, reflecting the degree of development of the bones, can be used to predict the adult height and detect endocrine diseases of children. Both examinations of radiologists and variability of operators have a significant impact on bone age assessment. To decrease human intervention , machine learning algorithms are used to assess the bone age automatically. However, conventional supervised deep-learning methods need pre-labeled data. In this paper, based on the convolutional auto-encoder with constraints (CCAE), an unsupervised deep-learning model proposed in the classification of the fingerprint, we propose this model for the classification of the bone age and baptize it BA-CCAE. In the proposed BA-CCAE model, the key regions of the raw X-ray images of the bone age are encoded, yielding the latent vectors. The K-means clustering algorithm is used to obtain the final classifications by grouping the latent vectors of the bone images. A set of experiments on the Radiological Society of North America pediatric bone age dataset (RSNA) show that the accuracy of classifications at 48-month intervals is 76.15%. Although the accuracy now is lower than most of the existing supervised models, the proposed BA-CCAE model can establish the classification of bone age without any pre-labeled data, and to the best of our knowledge, the proposed BA-CCAE is one of the few trails using the unsupervised deep-learning method for the bone age assessment.
Early prostate cancer detection and staging from MRI are extremely challenging tasks for both radiologists and deep learning algorithms, but the potential to learn from large and diverse datasets remains a promising avenue to increase their generalization capability both within- and across clinics. To enable this for prototype-stage algorithms, where the majority of existing research remains, in this paper we introduce a flexible federated learning framework for cross-site training, validation, and evaluation of deep prostate cancer detection algorithms. Our approach utilizes an abstracted representation of the model architecture and data, which allows unpolished prototype deep learning models to be trained without modification using the NVFlare federated learning framework. Our results show increases in prostate cancer detection and classification accuracy using a specialized neural network model and diverse prostate biopsy data collected at two University of California research hospitals, demonstrating the efficacy of our approach in adapting to different datasets and improving MR-biomarker discovery. We open-source our FLtools system, which can be easily adapted to other deep learning projects for medical imaging.
In recent years, deep learning has achieved remarkable empirical success for image reconstruction. This has catalyzed an ongoing quest for precise characterization of correctness and reliability of data-driven methods in critical use-cases, for instance in medical imaging. Notwithstanding the excellent performance and efficacy of deep learning-based methods, concerns have been raised regarding their stability, or lack thereof, with serious practical implications. Significant advances have been made in recent years to unravel the inner workings of data-driven image recovery methods, challenging their widely perceived black-box nature. In this article, we will specify relevant notions of convergence for data-driven image reconstruction, which will form the basis of a survey of learned methods with mathematically rigorous reconstruction guarantees. An example that is highlighted is the role of ICNN, offering the possibility to combine the power of deep learning with classical convex regularization theory for devising methods that are provably convergent. This survey article is aimed at both methodological researchers seeking to advance the frontiers of our understanding of data-driven image reconstruction methods as well as practitioners, by providing an accessible description of useful convergence concepts and by placing some of the existing empirical practices on a solid mathematical foundation.
Subject motion in whole-body dynamic PET introduces inter-frame mismatch and seriously impacts parametric imaging. Traditional non-rigid registration methods are generally computationally intense and time-consuming. Deep learning approaches are promising in achieving high accuracy with fast speed, but have yet been investigated with consideration for tracer distribution changes or in the whole-body scope. In this work, we developed an unsupervised automatic deep learning-based framework to correct inter-frame body motion. The motion estimation network is a convolutional neural network with a combined convolutional long short-term memory layer, fully utilizing dynamic temporal features and spatial information. Our dataset contains 27 subjects each under a 90-min FDG whole-body dynamic PET scan. With 9-fold cross-validation, compared with both traditional and deep learning baselines, we demonstrated that the proposed network obtained superior performance in enhanced qualitative and quantitative spatial alignment between parametric $K_{i}$ and $V_{b}$ images and in significantly reduced parametric fitting error. We also showed the potential of the proposed motion correction method for impacting downstream analysis of the estimated parametric images, improving the ability to distinguish malignant from benign hypermetabolic regions of interest. Once trained, the motion estimation inference time of our proposed network was around 460 times faster than the conventional registration baseline, showing its potential to be easily applied in clinical settings.
In this paper, a novel solution is introduced for visual Simultaneous Localization and Mapping (vSLAM) that is built up of Deep Learning components. The proposed architecture is a highly modular framework in which each component offers state of the art results in their respective fields of vision-based deep learning solutions. The paper shows that with the synergic integration of these individual building blocks, a functioning and efficient all-through deep neural (ATDN) vSLAM system can be created. The Embedding Distance Loss function is introduced and using it the ATDN architecture is trained. The resulting system managed to achieve 4.4% translation and 0.0176 deg/m rotational error on a subset of the KITTI dataset. The proposed architecture can be used for efficient and low-latency autonomous driving (AD) aiding database creation as well as a basis for autonomous vehicle (AV) control.
Object detection and recognition are fundamental functions underlying the success of species. Because the appearance of an object exhibits a large variability, the brain has to group these different stimuli under the same object identity, a process of generalization. Does the process of generalization follow some general principles or is it an ad-hoc "bag-of-tricks"? The Universal Law of Generalization provided evidence that generalization follows similar properties across a variety of species and tasks. Here we test the hypothesis that the internal representations underlying generalization reflect the natural properties of object detection and recognition in our environment rather than the specifics of the system solving these problems. By training a deep-neural-network with images of "clear" and "camouflaged" animals, we found that with a proper choice of category prototypes, the generalization functions are monotone decreasing, similar to the generalization functions of biological systems. Our findings support the hypothesis of the study.
Magnetic resonance (MR) image re-parameterization refers to the process of generating via simulations of an MR image with a new set of MRI scanning parameters. Different parameter values generate distinct contrast between different tissues, helping identify pathologic tissue. Typically, more than one scan is required for diagnosis; however, acquiring repeated scans can be costly, time-consuming, and difficult for patients. Thus, using MR image re-parameterization to predict and estimate the contrast in these imaging scans can be an effective alternative. In this work, we propose a novel deep learning (DL) based convolutional model for MRI re-parameterization. Based on our preliminary results, DL-based techniques hold the potential to learn the non-linearities that govern the re-parameterization.
By adding exiting layers to the deep learning networks, early exit can terminate the inference earlier with accurate results. The passive decision-making of whether to exit or continue the next layer has to go through every pre-placed exiting layer until it exits. In addition, it is also hard to adjust the configurations of the computing platforms alongside the inference proceeds. By incorporating a low-cost prediction engine, we propose a Predictive Exit framework for computation- and energy-efficient deep learning applications. Predictive Exit can forecast where the network will exit (i.e., establish the number of remaining layers to finish the inference), which effectively reduces the network computation cost by exiting on time without running every pre-placed exiting layer. Moreover, according to the number of remaining layers, proper computing configurations (i.e., frequency and voltage) are selected to execute the network to further save energy. Extensive experimental results demonstrate that Predictive Exit achieves up to 96.2% computation reduction and 72.9% energy-saving compared with classic deep learning networks; and 12.8% computation reduction and 37.6% energy-saving compared with the early exit under state-of-the-art exiting strategies, given the same inference accuracy and latency.
Musical form analysis is a rigorous task that frequently challenges the expertise of human analysts and signal processing algorithms alike. While numerous systems have been proposed to perform the tasks of musical segmentation, genre classification, and single-label segment classification in popular music, none have specifically focused on the analytical process used by classical musicians. Classical music form analysis facilitates a combination of these tasks, including form classification, structural segmentation, and multilabel large- and small-segment classification – tasks that lack feasible algorithms, machine learning models, and extensive research. Form analysis has many applications in the world of music, and a viable analytical system would greatly benefit performing musicians and academic researchers, both in musicology and signal processing. As well, current datasets used for related research tasks lack standardized analytical conventions, including form classification, and suffer from erroneous annotations and extensibility due to the data sources used for the music. In this thesis, we propose a new system to perform the task of automatic musical form analysis using deep learning models, as well as a new standardized dataset.
In this thesis, we try to build a connection between the two schools by introducing syntactic inductive biases for deep learning models. We propose two families of inductive biases, one for constituency structure and another one for dependency structure. The constituency inductive bias encourages deep learning models to use different units (or neurons) to separately process long-term and short-term information. This separation provides a way for deep learning models to build the latent hierarchical representations from sequential inputs, that a higher-level representation is composed of and can be decomposed into a series of lower-level representations. For example, without knowing the ground-truth structure, our proposed model learns to process logical expression through composing representations of variables and operators into representations of expressions according to its syntactic structure. On the other hand, the dependency inductive bias encourages models to find the latent relations between entities in the input sequence. For natural language, the latent relations are usually modeled as a directed dependency graph, where a word has exactly one parent node and zero or several children nodes. After applying this constraint to a Transformer-like model, we find the model is capable of inducing directed graphs that are close to human expert annotations, and it also outperforms the standard transformer model on different tasks. We believe that these experimental results demonstrate an interesting alternative for the future development of deep learning models.