Choice modeling has been a central topic in the study of individual preference or utility across many fields including economics, marketing, operations research, and psychology. While the vast majority of the literature on choice models has been devoted to the analytical properties that lead to managerial and policy-making insights, the existing methods to learn a choice model from empirical data are often either computationally intractable or sample inefficient. In this paper, we develop deep learning-based choice models under two settings of choice modeling: (i) feature-free and (ii) feature-based. Our model captures both the intrinsic utility for each candidate choice and the effect that the assortment has on the choice probability. Synthetic and real data experiments demonstrate the performances of proposed models in terms of the recovery of the existing choice models, sample complexity, assortment effect, architecture design, and model interpretation.
Electrocardiogram (ECG) signals are beneficial in diagnosing cardiovascular diseases, which are one of the leading causes of death. However, they are often contaminated by noise artifacts and affect the automatic and manual diagnosis process. Automatic deep learning-based examination of ECG signals can lead to inaccurate diagnosis, and manual analysis involves rejection of noisy ECG samples by clinicians, which might cost extra time. To address this limitation, we present a two-stage deep learning-based framework to automatically detect the noisy ECG samples. Through extensive experiments and analysis on two different datasets, we observe that the deep learning-based framework can detect slightly and highly noisy ECG samples effectively. We also study the transfer of the model learned on one dataset to another dataset and observe that the framework effectively detects noisy ECG samples.
Simulation of turbulent flows, especially at the edges of clouds in the atmosphere, is an inherently challenging task. Hitherto, the best possible computational method to perform such experiments is the Direct Numerical Simulation (DNS). DNS involves solving non-linear partial differential equations for fluid flows, also known as Navier-Stokes equations, on discretized grid boxes in a three-dimensional space. It is a valuable paradigm that has guided the numerical weather prediction models to compute rainfall formation. However, DNS cannot be performed for large domains of practical utility to the weather forecast community. Here, we introduce DeepClouds.ai, a 3D-UNET that simulates the outputs of a rising cloud DNS experiment. The problem of increasing the domain size in DNS is addressed by mapping an inner 3D cube to the complete 3D cube from the output of the DNS discretized grid simulation. Our approach effectively captures turbulent flow dynamics without having to solve the complex dynamical core. The baseline shows that the deep learning-based simulation is comparable to the partial-differential equation-based model as measured by various score metrics. This framework can be used to further the science of turbulence and cloud flows by enabling simulations over large physical domains in the atmosphere. It would lead to cascading societal benefits by improved weather predictions via advanced parameterization schemes.
In this paper, we investigate the transferability limitations when using deep learning models, for semantic segmentation of pneumonia-infected areas in CT images. The proposed approach adopts a 4 channel input; 3 channels based on Hounsfield scale, plus one channel (binary) denoting the lung area. We used 3 different, publicly available, CT datasets. If the lung area mask was not available, a deep learning model generates a proxy image. Experimental results suggesting that transferability should be used carefully, when creating Covid segmentation models; retraining the model more than one times in large sets of data results in a decrease in segmentation accuracy.
This paper presents a method for time-lapse 3D cell analysis. Specifically, we consider the problem of accurately localizing and quantitatively analyzing sub-cellular features, and for tracking individual cells from time-lapse 3D confocal cell image stacks. The heterogeneity of cells and the volume of multi-dimensional images presents a major challenge for fully automated analysis of morphogenesis and development of cells. This paper is motivated by the pavement cell growth process, and building a quantitative morphogenesis model. We propose a deep feature based segmentation method to accurately detect and label each cell region. An adjacency graph based method is used to extract sub-cellular features of the segmented cells. Finally, the robust graph based tracking algorithm using multiple cell features is proposed for associating cells at different time instances. Extensive experiment results are provided and demonstrate the robustness of the proposed method. The code is available on Github and the method is available as a service through the BisQue portal.
Metal artifact correction is a challenging problem in cone beam computed tomography (CBCT) scanning. Metal implants inserted into the anatomy cause severe artifacts in reconstructed images. Widely used inpainting-based metal artifact reduction (MAR) methods require segmentation of metal traces in the projections as a first step, which is a challenging task. One approach is to use a deep learning method to segment metals in the projections. However, the success of deep learning methods is limited by the availability of realistic training data. It is laborious and time consuming to get reliable ground truth annotations due to unclear implant boundaries and large numbers of projections. We propose to use X-ray simulations to generate synthetic metal segmentation training dataset from clinical CBCT scans. We compare the effect of simulations with different numbers of photons and also compare several training strategies to augment the available data. We compare our model's performance on real clinical scans with conventional region growing threshold-based MAR, moving metal artifact reduction method, and a recent deep learning method. We show that simulations with relatively small number of photons are suitable for the metal segmentation task and that training the deep learning model with full size and cropped projections together improves the robustness of the model. We show substantial improvement in the image quality affected by severe motion, voxel size under-sampling, and out-of-FOV metals. Our method can be easily integrated into the existing projection-based MAR pipeline to get improved image quality. This method can provide a novel paradigm to accurately segment metals in CBCT projections.
Deep neural networks for survival prediction outper-form classical approaches in discrimination, which is the ordering of patients according to their time-of-event. Conversely, classical approaches like the Cox Proportional Hazards model display much better calibration, the correct temporal prediction of events of the underlying distribution. Especially in the medical domain, where it is critical to predict the survival of a single patient, both discrimination and calibration are important performance metrics. Here we present Discrete Calibrated Survival (DCS), a novel deep neural network for discriminated and calibrated survival prediction that outperforms competing survival models in discrimination on three medical datasets, while achieving best calibration among all discrete time models. The enhanced performance of DCS can be attributed to two novel features, the variable temporal output node spacing and the novel loss term that optimizes the use of uncensored and censored patient data. We believe that DCS is an important step towards clinical application of deep-learning-based survival prediction with state-of-the-art discrimination and good calibration.
Atmospheric effects, such as turbulence and background thermal noise, inhibit the propagation of coherent light used in ON-OFF keying free-space optical communication. Here we present and experimentally validate a convolutional neural network to reduce the bit error rate of free-space optical communication in post-processing that is significantly simpler and cheaper than existing solutions based on advanced optics. Our approach consists of two neural networks, the first determining the presence of coherent bit sequences in thermal noise and turbulence and the second demodulating the coherent bit sequences. All data used for training and testing our network is obtained experimentally by generating ON-OFF keying bit streams of coherent light, combining these with thermal light, and passing the resultant light through a turbulent water tank which we have verified mimics turbulence in the air to a high degree of accuracy. Our convolutional neural network improves detection accuracy over threshold classification schemes and has the capability to be integrated with current demodulation and error correction schemes.
The convergence of 5G architecture and deep learning has gained a lot of research interests in both the fields of wireless communication and artificial intelligence. This is because deep learning technologies have been identified to be the potential driver of the 5G technologies, that make up the 5G architecture. Hence, there have been extensive surveys on the convergence of 5G architecture and deep learning. However, most of the existing survey papers mainly focused on how deep learning can converge with a specific 5G technology, thus, not covering the full spectrum of the 5G architecture. Although there is a recent survey paper that appears to be robust, a review of that paper shows that it is not well structured to specifically cover the convergence of deep learning and the 5G technologies. Hence, this paper provides a robust overview of the convergence of the key 5G technologies and deep learning. The challenges faced by such convergence are discussed. In addition, a brief overview of the future 6G architecture, and how it can converge with deep learning is also discussed.
Accurate sleep stage classification is significant for sleep health assessment. In recent years, several machine-learning based sleep staging algorithms have been developed , and in particular, deep-learning based algorithms have achieved performance on par with human annotation. Despite improved performance, a limitation of most deep-learning based algorithms is their black-box behavior, which have limited their use in clinical settings. Here, we propose a cross-modal transformer, which is a transformer-based method for sleep stage classification. The proposed cross-modal transformer consists of a novel cross-modal transformer encoder architecture along with a multi-scale one-dimensional convolutional neural network for automatic representation learning. Our method outperforms the state-of-the-art methods and eliminates the black-box behavior of deep-learning models by utilizing the interpretability aspect of the attention modules. Furthermore, our method provides considerable reductions in the number of parameters and training time compared to the state-of-the-art methods. Our code is available at https://github.com/Jathurshan0330/Cross-Modal-Transformer. A demo of our work can be found at https://bit.ly/Cross_modal_transformer_demo.
In deep learning, different kinds of deep networks typically need different optimizers, which have to be chosen after multiple trials, making the training process inefficient. To relieve this issue and consistently improve the model training speed across deep networks, we propose the ADAptive Nesterov momentum algorithm, Adan for short. Adan first reformulates the vanilla Nesterov acceleration to develop a new Nesterov momentum estimation (NME) method, which avoids the extra overhead of computing gradient at the extrapolation point. Then, Adan adopts NME to estimate the gradient's first- and second-order moments in adaptive gradient algorithms for convergence acceleration. Besides, we prove that Adan finds an $\epsilon$-approximate first-order stationary point within $\mathcal{O}(\epsilon^{-3.5})$ stochastic gradient complexity on the non-convex stochastic problems (e.g., deep learning problems), matching the best-known lower bound. Extensive experimental results show that Adan consistently surpasses the corresponding SoTA optimizers on vision, language, and RL tasks and sets new SoTAs for many popular networks and frameworks, e.g., ResNet, ConvNext, ViT, Swin, MAE, DETR, GPT-2, Transformer-XL, and BERT. More surprisingly, Adan can use half of the training cost (epochs) of SoTA optimizers to achieve higher or comparable performance on ViT, GPT-2, MAE, etc., and also shows great tolerance to a large range of minibatch size, e.g., from 1k to 32k. Code is released at https://github.com/sail-sg/Adan, and has been used in multiple popular deep learning frameworks or projects.
High energy density physics (HEDP) experiments commonly involve a dynamic wave-front propagating inside a low-density foam. This effect affects its density and hence, its transparency. A common problem in foam production is the creation of defective foams. Accurate information on their dimension and homogeneity is required to classify the foams' quality. Therefore, those parameters are being characterized using a 3D-measuring laser confocal microscope. For each foam, five images are taken: two 2D images representing the top and bottom surface foam planes and three images of side cross-sections from 3D scannings. An expert has to do the complicated, harsh, and exhausting work of manually classifying the foam's quality through the image set and only then determine whether the foam can be used in experiments or not. Currently, quality has two binary levels of normal vs. defective. At the same time, experts are commonly required to classify a sub-class of normal-defective, i.e., foams that are defective but might be sufficient for the needed experiment. This sub-class is problematic due to inconclusive judgment that is primarily intuitive. In this work, we present a novel state-of-the-art multi-view deep learning classification model that mimics the physicist's perspective by automatically determining the foams' quality classification and thus aids the expert. Our model achieved 86\% accuracy on upper and lower surface foam planes and 82\% on the entire set, suggesting interesting heuristics to the problem. A significant added value in this work is the ability to regress the foam quality instead of binary deduction and even explain the decision visually. The source code used in this work, as well as other relevant sources, are available at: https://github.com/Scientific-Computing-Lab-NRCN/Multi-View-Foams.git
Measurement of the cardiac pulse from facial video has become an interesting pursuit of research over the last few years. This is mainly due to the increasing importance of obtaining the heart rate of an individual in a non-invasive manner, which can be highly useful for applications in gaming and the medical industry. Another instrumental area of research over the past few years has been the advent of Deep Learning and using Deep Neural networks to enhance task performance. In this work, we propose to use efficient convolutional networks to accurately measure the heart rate of user from low resolution facial videos. Furthermore, to ensure that we are able to obtain the heart rate in real time, we compress the deep learning model by pruning it, thereby reducing its memory footprint. We benchmark the performance of our approach on the MAHNOB dataset and compare its performance across multiple approaches.
Object detection systems using deep learning models have become increasingly popular in robotics thanks to the rising power of CPUs and GPUs in embedded systems. However, these models are susceptible to adversarial attacks. While some attacks are limited by strict assumptions on access to the detection system, we propose a novel hardware attack inspired by Man-in-the-Middle attacks in cryptography. This attack generates a Universal Adversarial Perturbations (UAP) and injects the perturbation between the USB camera and the detection system via a hardware attack. Besides, prior research is misled by an evaluation metric that measures the model accuracy rather than the attack performance. In combination with our proposed evaluation metrics, we significantly increased the strength of adversarial perturbations. These findings raise serious concerns for applications of deep learning models in safety-critical systems, such as autonomous driving.
Automatic detection of traffic accidents is an important emerging topic in traffic monitoring systems. Nowadays many urban intersections are equipped with surveillance cameras connected to traffic management systems. Therefore, computer vision techniques can be viable tools for automatic accident detection. This paper presents a new efficient framework for accident detection at intersections for traffic surveillance applications. The proposed framework consists of three hierarchical steps, including efficient and accurate object detection based on the state-of-the-art YOLOv4 method, object tracking based on Kalman filter coupled with the Hungarian algorithm for association, and accident detection by trajectory conflict analysis. A new cost function is applied for object association to accommodate for occlusion, overlapping objects, and shape changes in the object tracking step. The object trajectories are analyzed in terms of velocity, angle, and distance in order to detect different types of trajectory conflicts including vehicle-to-vehicle, vehicle-to-pedestrian, and vehicle-to-bicycle. Experimental results using real traffic video data show the feasibility of the proposed method in real-time applications of traffic surveillance. In particular, trajectory conflicts, including near-accidents and accidents occurring at urban intersections are detected with a low false alarm rate and a high detection rate. The robustness of the proposed framework is evaluated using video sequences collected from YouTube with diverse illumination conditions. The dataset is publicly available at: http://github.com/hadi-ghnd/AccidentDetection.
The availability of reliable, high-resolution climate and weather data is important to inform long-term decisions on climate adaptation and mitigation and to guide rapid responses to extreme events. Forecasting models are limited by computational costs and, therefore, often generate coarse-resolution predictions. Statistical downscaling, including super-resolution methods from deep learning, can provide an efficient method of upsampling low-resolution data. However, despite achieving visually compelling results in some cases, such models frequently violate conservation laws when predicting physical variables. In order to conserve physical quantities, here we introduce methods that guarantee statistical constraints are satisfied by a deep learning downscaling model, while also improving their performance according to traditional metrics. We compare different constraining approaches and demonstrate their applicability across different neural architectures as well as a variety of climate and weather data sets. Besides enabling faster and more accurate climate predictions through downscaling, we also show that our novel methodologies can improve super-resolution for satellite data and natural images data sets.
Neural Network (Deep Learning) is a modern model in Artificial Intelligence and it has been exploited in Survival Analysis. Although several improvements have been shown by previous works, training an excellent deep learning model requires a huge amount of data, which may not hold in practice. To address this challenge, we develop a Kullback-Leibler-based (KL) deep learning procedure to integrate external survival prediction models with newly collected time-to-event data. Time-dependent KL discrimination information is utilized to measure the discrepancy between the external and internal data. This is the first work considering using prior information to deal with short data problem in Survival Analysis for deep learning. Simulation and real data results show that the proposed model achieves better performance and higher robustness compared with previous works.
Advancements in Telemedicine as an approach to healthcare delivery have heralded a new dawn in modern Medicine. Its fast-paced development in our contemporary society is credence to the advances in Artificial Intelligence and Information Technology. This paper carries out a descriptive study to broadly explore AI's implementations in healthcare delivery with a more holistic view of the usability of various Telemedical Innovations in enhancing Virtual Diagnostic Solutions (VDS). This research further explores notable developments in Deep Learning model optimizations for Virtual Diagnostic Solutions. A further research review on the prospects of Virtual Diagnostic Solutions (VDS) and foreseeable challenges was also highlighted. Conclusively, this research gives a general overview of Artificial Intelligence in Telemedicine with a central focus on Deep Learning-based approaches to Virtual Diagnostic Solutions.
This paper introduces a deep learning system based on a quantum neural network for the binary classification of points of a specific geometric pattern (Two-Moons Classification problem) on a plane. We believe that the use of hybrid deep learning systems (classical + quantum) can reasonably bring benefits, not only in terms of computational acceleration but in understanding the underlying phenomena and mechanisms; that will lead to the creation of new forms of machine learning, as well as to a strong development in the world of quantum computation. The chosen dataset is based on a 2D binary classification generator, which helps test the effectiveness of specific algorithms; it is a set of 2D points forming two interspersed semicircles. It displays two disjointed data sets in a two-dimensional representation space: the features are, therefore, the individual points' two coordinates, $x_1$ and $x_2$. The intention was to produce a quantum deep neural network with the minimum number of trainable parameters capable of correctly recognising and classifying points.
Photon-counting CT (PCCT) offers improved diagnostic performance through better spatial and energy resolution, but developing high-quality image reconstruction methods that can deal with these large datasets is challenging. Model-based solutions incorporate models of the physical acquisition in order to reconstruct more accurate images, but are dependent on an accurate forward operator and present difficulties with finding good regularization. Another approach is deep-learning reconstruction, which has shown great promise in CT. However, fully data-driven solutions typically need large amounts of training data and lack interpretability. To combine the benefits of both methods, while minimizing their respective drawbacks, it is desirable to develop reconstruction algorithms that combine both model-based and data-driven approaches. In this work, we present a novel deep-learning solution for material decomposition in PCCT, based on an unrolled/unfolded iterative network. We evaluate two cases: a learned post-processing, which implicitly utilizes model knowledge, and a learned gradient-descent, which has explicit model-based components in the architecture. With our proposed techniques, we solve a challenging PCCT simulation case: three-material decomposition in abdomen imaging with low dose, iodine contrast, and a very small training sample support. In this scenario, our approach outperforms a maximum likelihood estimation, a variational method, as well as a fully-learned network.