Nowadays, diseases are increasing in numbers and severity by the hour. Immunity diseases, affecting 8\% of the world population in 2017 according to the World Health Organization (WHO), is a field in medicine worth attention due to the high rate of disease occurrence classified under this category. This work presents an up-to-date review of state-of-the-art immune diseases healthcare solutions. We focus on tackling the issue with modern solutions such as Deep Learning to detect anomalies in the early stages hence providing health practitioners with efficient tools. We rely on advanced deep learning techniques such as Convolutional Neural Networks (CNN) to fulfill our objective of providing an efficient tool while providing a proficient analysis of this solution. The proposed solution was tested and evaluated by the immunology department in the Principal Military Hospital of Instruction of Tunis, which considered it a very helpful tool.
In this paper, we address the problem of direction of arrival (DOA) estimation for multiple targets in the presence of sensor failures in a sparse array. Generally, sparse arrays are known with very high-resolution capabilities, where N physical sensors can resolve up to $\mathcal{O}(N^2)$ uncorrelated sources. However, among the many configurations introduced in the literature, the arrays that provide the largest hole-free co-array are the most susceptible to sensor failures. We propose here two machine learning (ML) methods to mitigate the effect of sensor failures and maintain the DOA estimation performance and resolution. The first method enhances the conventional spatial smoothing using deep neural network (DNN), while the second one is an end-to-end data-driven method. Numerical results show that both approaches can significantly improve the performance of MRA with two failed sensors. The data-driven method can maintain the performance of the array with no failures at high signal-tonoise ratio (SNR). Moreover, both approaches can even perform better than the original array at low SNR thanks to the denoising effect of the proposed DNN
Western countries rely heavily on wheat, and yield prediction is crucial. Time-series deep learning models, such as Long Short Term Memory (LSTM), have already been explored and applied to yield prediction. Existing literature reported that they perform better than traditional Machine Learning (ML) models. However, the existing LSTM cannot handle heterogeneous datasets (a combination of data which varies and remains static with time). In this paper, we propose an efficient deep learning model that can deal with heterogeneous datasets. We developed the system architecture and applied it to the real-world dataset in the digital agriculture area. We showed that it outperforms the existing ML models.
With the increase in the scale of Deep Learning (DL) training workloads in terms of compute resources and time consumption, the likelihood of encountering in-training failures rises substantially, leading to lost work and resource wastage. Such failures are typically offset by a checkpointing mechanism, which comes at the cost of storage and network bandwidth overhead. State-of-the-art approaches involve lossy model compression mechanisms, which induce a tradeoff between the resulting model quality (accuracy) and compression ratio. Delta compression is then also used to further reduce the overhead by only storing the difference between consecutive checkpoints. We make a key enabling observation that the sensitivity of model weights to compression varies during training, and different weights benefit from different quantization levels (ranging from retaining full precision to pruning). We propose (1) a non-uniform quantization scheme that leverages this variation, (2) an efficient search mechanism to dynamically adjust to the best quantization configurations, and (3) a quantization-aware delta compression mechanism that rearranges weights to minimize checkpoint differences, thereby maximizing compression. We instantiate these contributions in DynaQuant - a framework for DL workload checkpoint compression. Our experiments show that DynaQuant consistently achieves better tradeoff between accuracy and compression ratios compared to prior works, enabling a compression ratio up to 39x and withstanding up to 10 restores with negligible accuracy impact for fault-tolerant training. DynaQuant achieves at least an order of magnitude reduction in checkpoint storage overhead for training failure recovery as well as transfer learning use cases without any loss of accuracy
Structure-based drug design (SBDD) leverages the three-dimensional geometry of proteins to identify potential drug candidates. Traditional approaches, rooted in physicochemical modeling and domain expertise, are often resource-intensive. Recent advancements in geometric deep learning, which effectively integrate and process 3D geometric data, alongside breakthroughs in accurate protein structure predictions from tools like AlphaFold, have significantly propelled the field forward. This paper systematically reviews the state-of-the-art in geometric deep learning for SBDD. We begin by outlining foundational tasks in SBDD, discussing prevalent 3D protein representations, and highlighting representative predictive and generative models. Next, we provide an in-depth review of key tasks, including binding site prediction, binding pose generation, de novo molecule generation, linker design, protein pocket generation, and binding affinity prediction. For each task, we present formal problem definitions, key methods, datasets, evaluation metrics, and performance benchmarks. Lastly, we explore current challenges and future opportunities in SBDD. Challenges include oversimplified problem formulations, limited out-of-distribution generalization, biosecurity concerns related to the misuse of structural data, insufficient evaluation metrics and large-scale benchmarks, and the need for experimental validation and enhanced model interpretability. Opportunities lie in leveraging multimodal datasets, integrating domain knowledge, developing comprehensive benchmarks, establishing criteria aligned with clinical outcomes, and designing foundation models to expand the scope of design tasks. We also curate \url{https://github.com/zaixizhang/Awesome-SBDD}, reflecting ongoing contributions and new datasets in SBDD.
The perception module of self-driving vehicles relies on a multi-sensor system to understand its environment. Recent advancements in deep learning have led to the rapid development of approaches that integrate multi-sensory measurements to enhance perception capabilities. This paper surveys the latest deep learning integration techniques applied to the perception module in autonomous driving systems, categorizing integration approaches based on "what, how, and when to integrate". A new taxonomy of integration is proposed, based on three dimensions: multi-view, multi-modality, and multi-frame. The integration operations and their pros and cons are summarized, providing new insights into the properties of an "ideal" data integration approach that can alleviate the limitations of existing methods. After reviewing hundreds of relevant papers, this survey concludes with a discussion of the key features of an optimal data integration approach.
Ground-based solar image restoration is a computationally expensive procedure that involves nonlinear optimization techniques. The presence of atmospheric turbulence produces perturbations in individual images that make it necessary to apply blind deconvolution techniques. These techniques rely on the observation of many short exposure frames that are used to simultaneously infer the instantaneous state of the atmosphere and the unperturbed object. We have recently explored the use of machine learning to accelerate this process, with promising results. We build upon this previous work to propose several interesting improvements that lead to better models. As well, we propose a new method to accelerate the restoration based on algorithm unrolling. In this method, the image restoration problem is solved with a gradient descent method that is unrolled and accelerated aided by a few small neural networks. The role of the neural networks is to correct the estimation of the solution at each iterative step. The model is trained to perform the optimization in a small fixed number of steps with a curated dataset. Our findings demonstrate that both methods significantly reduce the restoration time compared to the standard optimization procedure. Furthermore, we showcase that these models can be trained in an unsupervised manner using observed images from three different instruments. Remarkably, they also exhibit robust generalization capabilities when applied to new datasets. To foster further research and collaboration, we openly provide the trained models, along with the corresponding training and evaluation code, as well as the training dataset, to the scientific community.
The reliability of deep learning accelerators (DLAs) used in autonomous driving systems has significant impact on the system safety. However, the DLA reliability is usually evaluated with low-level metrics like mean square errors of the output which remains rather different from the high-level metrics like total distance traveled before failure in autonomous driving. As a result, the high-level reliability metrics evaluated at the post-silicon stage may still lead to DLA design revision and result in expensive reliable DLA design iterations targeting at autonomous driving. To address the problem, we proposed a DLA-in-loop reliability evaluation platform to enable system reliability evaluation at the early DLA design stage.
With the continuous maturation and expansion of neural network technology, deep neural networks have been widely utilized as the fundamental building blocks of deep learning in a variety of applications, including speech recognition, machine translation, image processing, and the creation of recommendation systems. Therefore, many real-world complex problems can be solved by the deep learning techniques. As is known, traditional news recommendation systems mostly employ techniques based on collaborative filtering and deep learning, but the performance of these algorithms is constrained by the sparsity of the data and the scalability of the approaches. In this paper, we propose a recommendation model using granular neural network model to recommend news to appropriate channels by analyzing the properties of news. Specifically, a specified neural network serves as the foundation for the granular neural network that the model is considered to be build. Different information granularities are attributed to various types of news material, and different information granularities are released between networks in various ways. When processing data, granular output is created, which is compared to the interval values pre-set on various platforms and used to quantify the analysis's effectiveness. The analysis results could help the media to match the proper news in depth, maximize the public attention of the news and the utilization of media resources.
Hand gesture recognition based on surface electromyographic (sEMG) signals is a promising approach for developing Human-Machine Interfaces (HMIs) with a natural control, such as intuitive robot interfaces or poly-articulated prostheses. However, real-world applications are limited by reliability problems due to motion artefacts, postural and temporal variability, and sensor re-positioning. This master thesis is the first application of deep learning on the Unibo-INAIL dataset, the first public sEMG dataset exploring the variability between subjects, sessions and arm postures by collecting data over 8 sessions of each of 7 able-bodied subjects executing 6 hand gestures in 4 arm postures. Recent studies address variability with strategies based on training set composition, which improve inter-posture and inter-day generalization of non-deep machine learning classifiers, among which the RBF-kernel SVM yields the highest accuracy. The deep architecture realized in this work is a 1d-CNN inspired by a 2d-CNN reported to perform well on other public benchmark databases. On this 1d-CNN, various training strategies based on training set composition were implemented and tested. Multi-session training proves to yield higher inter-session validation accuracies than single-session training. Two-posture training proves the best postural training (proving the benefit of training on more than one posture) and yields 81.2% inter-posture test accuracy. Five-day training proves the best multi-day training, yielding 75.9% inter-day test accuracy. All results are close to the baseline. Moreover, the results of multi-day training highlight the phenomenon of user adaptation, indicating that training should also prioritize recent data. Though not better than the baseline, the achieved classification accuracies rightfully place the 1d-CNN among the candidates for further research.
The analysis of multivariate time series data is challenging due to the various frequencies of signal changes that can occur over both short and long terms. Furthermore, standard deep learning models are often unsuitable for such datasets, as signals are typically sampled at different rates. To address these issues, we introduce MultiWave, a novel framework that enhances deep learning time series models by incorporating components that operate at the intrinsic frequencies of signals. MultiWave uses wavelets to decompose each signal into subsignals of varying frequencies and groups them into frequency bands. Each frequency band is handled by a different component of our model. A gating mechanism combines the output of the components to produce sparse models that use only specific signals at specific frequencies. Our experiments demonstrate that MultiWave accurately identifies informative frequency bands and improves the performance of various deep learning models, including LSTM, Transformer, and CNN-based models, for a wide range of applications. It attains top performance in stress and affect detection from wearables. It also increases the AUC of the best-performing model by 5% for in-hospital COVID-19 mortality prediction from patient blood samples and for human activity recognition from accelerometer and gyroscope data. We show that MultiWave consistently identifies critical features and their frequency components, thus providing valuable insights into the applications studied.
Recently deep learning and machine learning approaches have been widely employed for various applications in acoustics. Nonetheless, in the area of sound field processing and reconstruction classic methods based on the solutions of wave equation are still widespread. Recently, physics-informed neural networks have been proposed as a deep learning paradigm for solving partial differential equations which govern physical phenomena, bridging the gap between purely data-driven and model based methods. Here, we exploit physics-informed neural networks to reconstruct the early part of missing room impulse responses in an uniform linear array. This methodology allows us to exploit the underlying law of acoustics, i.e., the wave equation, forcing the neural network to generate physically meaningful solutions given only a limited number of data points. The results on real measurements show that the proposed model achieves accurate reconstruction and performance in line with respect to state-of-the-art deep-learning and compress sensing techniques while maintaining a lightweight architecture.
Deep learning models have gained popularity in climate science, following their success in computer vision and other domains. For instance, researchers are increasingly employing deep learning techniques for downscaling climate data, drawing inspiration from image super-resolution models. However, there are notable differences between image data and climate data. While image data typically falls within a specific range (e.g., [0, 255]) and exhibits a relatively uniform or normal distribution, climate data can possess arbitrary value ranges and highly uneven distributions, such as precipitation data. This non-uniform distribution presents challenges when attempting to directly apply existing computer vision models to climate science tasks. Few studies have addressed this issue thus far. In this study, we explore the effects of loss functions and non-linear data pre-processing methods for deep learning models in the context of climate downscaling. We employ a climate downscaling experiment as an example to evaluate these factors. Our findings reveal that L1 loss and L2 loss perform similarly on some more balanced data like temperature data while for some imbalanced data like precipitation data, L2 loss performs significantly better than L1 loss. Additionally, we propose an approach to automatically learn the non-linear pre-processing function, which further enhances model accuracy and achieves the best results.
This thesis explored applications of the new emerging techniques of artificial intelligence and deep learning (neural networks in particular) for predictive maintenance, diagnostics and prognostics. Many neural architectures such as fully-connected, convolutional and recurrent neural networks were developed and tested on public datasets such as NASA C-MAPSS, Case Western Reserve University Bearings and FEMTO Bearings datasets to diagnose equipment health state and/or predict the remaining useful life (RUL) before breakdown. Many data processing and feature extraction procedures were used in combination with deep learning techniques such as dimensionality reduction (Principal Component Analysis) and signal processing (Fourier and Wavelet analyses) in order to create more meaningful and robust features to use as an input for neural networks architectures. This thesis also explored the potential use of these techniques in predictive maintenance within oil rigs for monitoring oilfield critical equipment in order to reduce unpredicted downtime and maintenance costs.
Black spot identification, a spatiotemporal phenomenon, involves analyzing the geographical location and time-based occurrence of road accidents. Typically, this analysis examines specific locations on road networks during set time periods to pinpoint areas with a higher concentration of accidents, known as black spots. By evaluating these problem areas, researchers can uncover the underlying causes and reasons for increased collision rates, such as road design, traffic volume, driver behavior, weather, and infrastructure. However, challenges in identifying black spots include limited data availability, data quality, and assessing contributing factors. Additionally, evolving road design, infrastructure, and vehicle safety technology can affect black spot analysis and determination. This study focused on traffic accidents in Greek road networks to recognize black spots, utilizing data from police and government-issued car crash reports. The study produced a publicly available dataset called Black Spots of North Greece (BSNG) and a highly accurate identification method.
In this paper, we introduce AdaSelection, an adaptive sub-sampling method to identify the most informative sub-samples within each minibatch to speed up the training of large-scale deep learning models without sacrificing model performance. Our method is able to flexibly combines an arbitrary number of baseline sub-sampling methods incorporating the method-level importance and intra-method sample-level importance at each iteration. The standard practice of ad-hoc sampling often leads to continuous training with vast amounts of data from production environments. To improve the selection of data instances during forward and backward passes, we propose recording a constant amount of information per instance from these passes. We demonstrate the effectiveness of our method by testing it across various types of inputs and tasks, including the classification tasks on both image and language datasets, as well as regression tasks. Compared with industry-standard baselines, AdaSelection consistently displays superior performance.
The integration of deep learning and neuroscience has been advancing rapidly, which has led to improvements in the analysis of brain activity and the understanding of deep learning models from a neuroscientific perspective. The reconstruction of visual experience from human brain activity is an area that has particularly benefited: the use of deep learning models trained on large amounts of natural images has greatly improved its quality, and approaches that combine the diverse information contained in visual experiences have proliferated rapidly in recent years. In this technical paper, by taking advantage of the simple and generic framework that we proposed (Takagi and Nishimoto, CVPR 2023), we examine the extent to which various additional decoding techniques affect the performance of visual experience reconstruction. Specifically, we combined our earlier work with the following three techniques: using decoded text from brain activity, nonlinear optimization for structural image reconstruction, and using decoded depth information from brain activity. We confirmed that these techniques contributed to improving accuracy over the baseline. We also discuss what researchers should consider when performing visual reconstruction using deep generative models trained on large datasets. Please check our webpage at https://sites.google.com/view/stablediffusion-with-brain/. Code is also available at https://github.com/yu-takagi/StableDiffusionReconstruction.
In recent years, there has been an increase in using deep learning for climate and weather modeling. Though results have been impressive, explainability and interpretability of deep learning models are still a challenge. A third wave of Artificial Intelligence (AI), which includes logic and reasoning, has been described as a way to address these issues. Neuro-symbolic AI is a key component of this integration of logic and reasoning with deep learning. In this work we propose a neuro-symbolic approach called Neuro-Symbolic Question-Answer Program Translator, or NS-QAPT, to address explainability and interpretability for deep learning climate simulation, applied to climate tipping point discovery. The NS-QAPT method includes a bidirectional encoder-decoder architecture that translates between domain-specific questions and executable programs used to direct the climate simulation, acting as a bridge between climate scientists and deep learning models. We show early compelling results of this translation method and introduce a domain-specific language and associated executable programs for a commonly known tipping point, the collapse of the Atlantic Meridional Overturning Circulation (AMOC).
Understanding Origin-Destination (O-D) travel demand is crucial for transportation management. However, traditional spatial-temporal deep learning models grapple with addressing the sparse and long-tail characteristics in high-resolution O-D matrices and quantifying prediction uncertainty. This dilemma arises from the numerous zeros and over-dispersed demand patterns within these matrices, which challenge the Gaussian assumption inherent to deterministic deep learning models. To address these challenges, we propose a novel approach: the Spatial-Temporal Tweedie Graph Neural Network (STTD). The STTD introduces the Tweedie distribution as a compelling alternative to the traditional 'zero-inflated' model and leverages spatial and temporal embeddings to parameterize travel demand distributions. Our evaluations using real-world datasets highlight STTD's superiority in providing accurate predictions and precise confidence intervals, particularly in high-resolution scenarios.
Objective: Magnetic Resonance Spectroscopy (MRS) is an important technique for biomedical detection. However, it is challenging to accurately quantify metabolites with proton MRS due to serious overlaps of metabolite signals, imperfections because of non-ideal acquisition conditions, and interference with strong background signals mainly from macromolecules. The most popular method, LCModel, adopts complicated non-linear least square to quantify metabolites and addresses these problems by designing empirical priors such as basis-sets, imperfection factors. However, when the signal-to-noise ratio of MRS signal is low, the solution may have large deviation. Methods: Linear Least Squares (LLS) is integrated with deep learning to reduce the complexity of solving this overall quantification. First, a neural network is designed to explicitly predict the imperfection factors and the overall signal from macromolecules. Then, metabolite quantification is solved analytically with the introduced LLS. In our Quantification Network (QNet), LLS takes part in the backpropagation of network training, which allows the feedback of the quantification error into metabolite spectrum estimation. This scheme greatly improves the generalization to metabolite concentrations unseen for training compared to the end-to-end deep learning method. Results: Experiments show that compared with LCModel, the proposed QNet, has smaller quantification errors for simulated data, and presents more stable quantification for 20 healthy in vivo data at a wide range of signal-to-noise ratio. QNet also outperforms other end-to-end deep learning methods. Conclusion: This study provides an intelligent, reliable and robust MRS quantification. Significance: QNet is the first LLS quantification aided by deep learning.