Application Of Deep Neural Networks Research Articles

Application of Deep Learning Neural Networks in Computer-aided Drug Discovery: A Review

Abstract: Computer-aided drug design has an important role in drug development and design. It has become a thriving area of research in the pharmaceutical industry to accelerate the drug discovery process. Deep learning, a subdivision of artificial intelligence, is widely applied to advance new drug development and design opportunities. This article reviews the recent technology that uses deep learning techniques to ameliorate the understanding of drug-target interactions in computer-aided drug discovery based on the prior knowledge acquired from various literature. In general, deep learning models can be trained to predict the binding affinity between the protein-ligand complexes and protein structures or generate protein-ligand complexes in structure-based drug discovery. In other words, artificial neural networks and deep learning algorithms, especially graph convolutional neural networks and generative adversarial networks, can be applied to drug discovery. Graph convolutional neural network effectively captures the interactions and structural information between atoms and molecules, which can be enforced to predict the binding affinity between protein and ligand. Also, the ligand molecules with the desired properties can be generated using generative adversarial networks.

Enhancing the Efficiency of Failure Recognition in Induction Machines through the Application of Deep Neural Networks

The objective of the investigation was to increase the effectiveness of damage detection in the stator of the squirrel-cage induction machine. The analysis aimed to enhance the operational trustworthiness of the squirrel-cage induction machine by employing nonintrusive diagnostic methods based on a current signal and modern artificial intelligence methods. The authors of the study introduced a diagnostic technique for identifying multiphase interturn short circuits of stator winding. These short circuits are one of the most common faults in induction machines. The proposed method focusses on deriving a diagnostic signal from the phase-current waveforms of the machine. The noninvasive nature of the diagnostic technique presented is attributed to the application of the field model of electromagnetic phenomena to determine the diagnostic signal. For this purpose, a field model of a squirrel-cage machine was developed. The waveforms of phase currents obtained from the field model were used as input into an elaborated machine failure neural classifier. A deep neural network was used to develop a neural classifier. The effectiveness of the developed classifier has been experimentally verified, and the obtained results have been presented, concluded, and discussed. The scientific novelty presented in the article is the presentation of research results on the use of a neural classifier to detect damage in all phases of the stator winding at an early stage of its appearance. The features of this type of damage are very difficult to observe in signal waveforms such as a phase current or torque.

PArtNNer: Platform-Agnostic Adaptive Edge-Cloud DNN Partitioning for Minimizing End-to-End Latency

The last decade has seen the emergence of Deep Neural Networks (DNNs) as the de facto algorithm for various computer vision applications. In intelligent edge devices, sensor data streams acquired by the device are processed by a DNN application running on either the edge device itself or in the cloud. However, “edge-only” and “cloud-only” execution of State-of-the-Art DNNs may not meet an application’s latency requirements due to the limited compute, memory, and energy resources in edge devices, dynamically varying bandwidth of edge-cloud connectivity networks, and temporal variations in the computational load of cloud servers. This work investigates distributed (partitioned) inference across edge devices (mobile/end device) and cloud servers to minimize end-to-end DNN inference latency. We study the impact of temporally varying operating conditions and the underlying compute and communication architecture on the decision of whether to run the inference solely on the edge, entirely in the cloud, or by partitioning the DNN model execution among the two. Leveraging the insights gained from this study and the wide variation in the capabilities of various edge platforms that run DNN inference, we propose PArtNNer , a platform-agnostic adaptive DNN partitioning algorithm that finds the optimal partitioning point in DNNs to minimize inference latency. PArtNNer can adapt to dynamic variations in communication bandwidth and cloud server load without requiring pre-characterization of underlying platforms. Experimental results for six image classification and object detection DNNs on a set of five commercial off-the-shelf compute platforms and three communication standards indicate that PArtNNer results in 10.2× and 3.2× (on average) and up to 21.1× and 6.7× improvements in end-to-end inference latency compared to execution of the DNN entirely on the edge device or entirely on a cloud server, respectively. Compared to pre-characterization-based partitioning approaches, PArtNNer converges to the optimal partitioning point 17.6× faster.

CMF defects database: A craniomaxillofacial defects dataset and a data-driven repair method

Due to the large number of medical images generated each year, data-driven machine learning methods hold great promise in the field of medical image diagnosis and analysis. In recent years, the application of deep neural networks in oral and cranio-maxillofacial surgery has gradually increased, helping surgeons to obtain more comprehensive information and develop more accurate treatment planning, including segmentation and recognition of anatomical structures, disease diagnosis, and automatic surgical planning. This paper presents a new dataset of maxillofacial defect images, containing computed tomography (CT) data of 125 normal human cranial skeletal 3D reconstructions and 2625 corresponding craniomaxillofacial bone defects. The data are classified according to seven different types of defects which are anterior skull and forehead, zygomatic arch, zygomatic bone, Naso-orbital-ethmoidal (NOE), wall of the maxillary sinus, mandibular body, mandibular ramus, and condylar respectively. Each kind of defect is further divided into three categories according to the different locations, including left side, right side, and bilateral sides (or defect passes through the midline of the face). There are three main contributions: first, a dataset is publicly available to assist researchers in the development of data-based methods for generating missing portions of the skull; second, a repair method is proposed to evaluate the dataset; third, a review of various related datasets. The release of this dataset aims to advance artificial intelligence methods in the field of oral and cranio-maxillofacial surgery and plastic surgery helping physicians to reduce the difficulty and time consumption while improving the efficiency and accuracy of prosthesis design.

An End-to-End Benchmarking Tool for Analyzing the Hardware-Software Implications of Multi-modal DNNs

Abstract-Multi-modal deep neural networks (DNNs) have become increasingly pervasive in many machine learning application domains due to their superior accuracy by fusing various modalities together. However, multi-modal DNNs present many unique characteristics such as multi-stage execution, frequent synchronization and high heterogeneity, which are not well understood in the system and architecture community. In this article, we first present and characterize a set of multi-modal DNN workloads of different sizes from five domains and measure metrics like accuracy to ensure the availability of these applications from the algorithm perspective. We then explore their important hardwaresoftware implications from system and architecture aspects by conducting an in-depth analysis on the unique hardware-software characteristics of multimodal DNNs. We hope that our work can help guide future hardware-software design and optimization for efficient inference of multi-modal DNN applications on both cloud and edge computing platforms.

HFTCRNet: Hierarchical Fusion Transformer for Interbank Credit Rating and Risk Assessment.

As a prominent application of deep neural networks in financial literature, bank credit ratings play a pivotal role in safeguarding global economic stability and preventing crises. In the contemporary financial system, interconnectivity among banks has reached unprecedented levels. However, many existing credit risk models continue to assess each bank independently, resulting in inevitable suboptimal performance. Thus, developing advanced neural networks to model intricate temporal dynamics and interconnected relationships in the banking system is essential for an effective credit rating and risk assessment learning system. To this end, we propose a novel hierarchical fusion transformer for interbank credit rating and risk assessment (HFTCRNet), which includes the long-term temporal transformer (LT 3 ) module, short-term cross-graph transformer (STCGT) module, attentive risk contagion transformer (ARCT) module, and hierarchical fusion transformer (HFT) module to capture the long-term growth trajectories of banks, the short-term interbank network variance, the potential propagation of risks within interbank network, and integrate these information hierarchically. We further develop an interbank credit rating dataset, encompassing quarterly financial data, interbank lending networks, and key indicators such as credit ratings and systemic risk (SRISK) for 4548 banks from 2016Q1 to 2023Q1. Notably, we also adapt the minimum density algorithm to stabilize the interbank loan network over time, aiding in the analysis of long-term and short-term network effects. Our learning system uses semi-supervised learning to handle labels of varying sparsity, integrating credit ratings and SRISK for a comprehensive assessment of individual bank creditworthiness and systemic interbank risk. Extensive experimental results on our interbank dataset show that HFTCRNet not only outperforms all the baselines in terms of credit rating accuracy but also can evaluate the systemic risk within the interbank network. Code will be available at: https://github.com/AI4Risk/HFTCRNet.

Research on Adversarial Sample Detection Method Based on Image Similarity

<p>With the widespread application of deep neural networks in image detection, adversarial sample attacks have gradually become a hot issue of concern for researchers. In this paper we propose a new adversarial sample detection approach called AdvDetector, which combines image generation through label fusion with image similarity detection. AdvDetector enhances sample quality and effectively identifies adversarial samples. Specifically, the method generates images by selecting seed pixels, the labels of deep neural network classification, and the pixel distribution learned from training data, and detects them using image similarity comparison methods. During the sample generation process, we introduce the AdvDetector method for adversarial sample detection to improve the quality of generated samples. We evaluated the effectiveness of the method on three publicly available image datasets, MNIST, Cifar-10, and GTSR, and the results show that the method is superior to existing baseline methods in terms of adversarial sample detection rate and sample generation quality.</p> <p> </p>

Efficient Application of Deep Neural Networks for Identifying Small and Multiple Weed Patches Using Drone Images

Efficient Application of Deep Neural Networks for Identifying Small and Multiple Weed Patches Using Drone Images

APPLICATION OF DEEP NEURAL NETWORK ON AUTONOMOUS UNDERWATER VEHICLES (AUV) IN SUBSEA PIPELINE INSPECTION

The purpose of this study is to investigate the application of deep neural network for the Automated Underwater Vehicle in the subsea pipeline inspection. In today’s modern world, we see more and more sophisticated and precise computer vision object detection technology being implemented in our daily lives. To name a few, security cameras, self-driving cars, drones and more. This research suggests that computer vision pipeline defect detection is an attractive solution for the future subsea pipeline defect detection as it relies less human intervention and is more reliable. In this paper, review of the current methods of subsea pipeline inspection was done. Some of the methods with visual detection and their limitations are also discussed. Apart from that, the machine learning algorithm of our focus, Faster RCNN is studied. Beyond that, we have explained the methods of our experimentation and the process of training as well as validating the custom dataset. The outcomes are separated into different sections, training curve, visual data representation as well as the model accuracy in different operating environment. In conclusion, we found that the model is able to detect the underwater pipeline with up to 1mm leak size. However, the low accuracy due to the insufficient dataset is recognized as a bottleneck and some of the recommendations are suggested for future improvement.

Mapping of Deep Neural Network Accelerators on Wireless Multistage Interconnection NoCs

In the last few decades, the concept of Wireless Network-on-chip (WiNoC) has emerged as a promising alternative for Multiprocessor Systems on Chip (MPSOC) to achieve reliable and scalable communication. Worth recalling in this regard is that our research team has already designed, verified and evaluated Multistage Interconnection Networks (MIN) in this field. With respect to the present work, we consider proceeding with further exploring our thoughts on this research area. Firstly, we propose the design and performance evaluation of a hybrid (wireless/wired) MIN, analysing how this augmented network can potentially improve not only the average delay, but also energy consumption. Secondly, we continue with examining the implementation of our advanced DELTA-based MIN architecture on Deep Neural Network (DNN) accelerators, while accounting for its potential regularity and scalability in simultaneously maintaining an effective power efficiency and lower latency throughout the DNN operating process. In this context, several metrics have been evaluated in regard to three DNN application cases through implementation of their main respective modules.

NeuroCool: Dynamic Thermal Management of 3D DRAM for Deep Neural Networks through Customized Prefetching

Deep neural network (DNN) implementations are typically characterized by huge datasets and concurrent computation, resulting in a demand for high memory bandwidth due to intensive data movement between processors and off-chip memory. Performing DNN inference on general-purpose cores/edge is gaining attraction to enhance user experience and reduce latency. The mismatch in the CPU and conventional DRAM speed leads to under-utilization of the compute capabilities, causing increased inference time. 3D DRAM is a promising solution to effectively fulfill the bandwidth requirement of high-throughput DNNs. However, due to high power density in stacked architectures, 3D DRAMs need dynamic thermal management (DTM), resulting in performance overhead due to memory-induced CPU throttling. We study the thermal impact of DNN applications running on a 3D DRAM system, and make a case for a memory temperature-aware customized prefetch mechanism to reduce DTM overheads and significantly improve performance. In our proposed NeuroCool DTM policy, we intelligently place either DRAM ranks or tiers in low power state, using the DNN layer characteristics and access rate. We establish the generalization of our approach through training and test datasets comprising diverse data points from widely used DNN applications. Experimental results on popular DNNs show that NeuroCool results in a average performance gain of 44% (as high as 52%) and memory energy improvement of 43% (as high as 69%) over general-purpose DTM policies.

Immunohistochemical Stain-Aided Annotation Accelerates Machine Learning and Deep Learning Model Development in the Pathologic Diagnosis of Nasopharyngeal Carcinoma.

Nasopharyngeal carcinoma (NPC) is an epithelial cancer originating in the nasopharynx epithelium. Nevertheless, annotating pathology slides remains a bottleneck in the development of AI-driven pathology models and applications. In the present study, we aim to demonstrate the feasibility of using immunohistochemistry (IHC) for annotation by non-pathologists and to develop an efficient model for distinguishing NPC without the time-consuming involvement of pathologists. For this study, we gathered NPC slides from 251 different patients, comprising hematoxylin and eosin (H&E) slides, pan-cytokeratin (Pan-CK) IHC slides, and Epstein-Barr virus-encoded small RNA (EBER) slides. The annotation of NPC regions in the H&E slides was carried out by a non-pathologist trainee who had access to corresponding Pan-CK IHC slides, both with and without EBER slides. The training process utilized ResNeXt, a deep neural network featuring a residual and inception architecture. In the validation set, NPC exhibited an AUC of 0.896, with a sensitivity of 0.919 and a specificity of 0.878. This study represents a significant breakthrough: the successful application of deep convolutional neural networks to identify NPC without the need for expert pathologist annotations. Our results underscore the potential of laboratory techniques to substantially reduce the workload of pathologists.

Computer-Aided Diagnosis of Laryngeal Cancer Based on Deep Learning with Laryngoscopic Images.

Laryngeal cancer poses a significant global health burden, with late-stage diagnoses contributing to reduced survival rates. This study explores the application of deep convolutional neural networks (DCNNs), specifically the Densenet201 architecture, in the computer-aided diagnosis of laryngeal cancer using laryngoscopic images. Our dataset comprised images from two medical centers, including benign and malignant cases, and was divided into training, internal validation, and external validation groups. We compared the performance of Densenet201 with other commonly used DCNN models and clinical assessments by experienced clinicians. Densenet201 exhibited outstanding performance, with an accuracy of 98.5% in the training cohort, 92.0% in the internal validation cohort, and 86.3% in the external validation cohort. The area under the curve (AUC) values consistently exceeded 92%, signifying robust discriminatory ability. Remarkably, Densenet201 achieved high sensitivity (98.9%) and specificity (98.2%) in the training cohort, ensuring accurate detection of both positive and negative cases. In contrast, other DCNN models displayed varying degrees of performance degradation in the external validation cohort, indicating the superiority of Densenet201. Moreover, Densenet201's performance was comparable to that of an experienced clinician (Clinician A) and outperformed another clinician (Clinician B), particularly in the external validation cohort. Statistical analysis, including the DeLong test, confirmed the significance of these performance differences. Our study demonstrates that Densenet201 is a highly accurate and reliable tool for the computer-aided diagnosis of laryngeal cancer based on laryngoscopic images. The findings underscore the potential of deep learning as a complementary tool for clinicians and the importance of incorporating advanced technology in improving diagnostic accuracy and patient care in laryngeal cancer diagnosis. Future work will involve expanding the dataset and further optimizing the deep learning model.

Application of deep neural survival networks to the development of risk prediction models for diabetes mellitus, hypertension, and dyslipidemia.

: Although numerous risk prediction models have been proposed, few such models have been developed using neural network-based survival analysis. We developed risk prediction models for three cardiovascular disease risk factors (diabetes mellitus, hypertension, and dyslipidemia) among a working-age population in Japan using DeepSurv, a deep feed-forward neural network. : Data were obtained from the Japan Epidemiology Collaboration on Occupational Health Study. A total of 51 258, 44 197, and 31 452 individuals were included in the development of risk models for diabetes mellitus, hypertension, and dyslipidemia, respectively; two-thirds of whom were used to develop prediction models, and the rest were used to validate the models. We compared the performances of DeepSurv-based models with those of prediction models based on the Cox proportional hazards model. : The area under the receiver-operating characteristic curve was 0.878 [95% confidence interval (CI) = 0.864-0.892] for diabetes mellitus, 0.835 (95% CI = 0.826-0.845) for hypertension, and 0.826 (95% CI = 0.817-0.835) for dyslipidemia. Compared with the Cox proportional hazards-based models, the DeepSurv-based models had better reclassification performance [diabetes mellitus: net reclassification improvement (NRI) = 0.474, P ≤ 0.001; hypertension: NRI = 0.194, P ≤ 0.001; dyslipidemia: NRI = 0.397, P ≤ 0.001] and discrimination performance [diabetes mellitus: integrated discrimination improvement (IDI) = 0.013, P ≤ 0.001; hypertension: IDI = 0.007, P ≤ 0.001; and dyslipidemia: IDI = 0.043, P ≤ 0.001]. : This study suggests that DeepSurv has the potential to improve the performance of risk prediction models for cardiovascular disease risk factors.

Application of Deep Neural Network and Human-Computer Interaction Technology Based on Multimodal Perception in Art Design

Abstract The essence of art design is the process of emotional interaction with humans. In this paper, emotions are classified using multimodal fusion features, and abstract fusion features are obtained by sampling the multimodal matching tensor using average pooling. The matching fusion matrix in the tensor operator is used to convert from two-modal to multi-modal matching. Virtual interaction model in art design controls the design objectives, and the optimized virtual world is constructed by using virtual reality technology, so as to build an immersive art design model. Finally, a study was conducted to examine the impact of the use of emotion perception and interaction technology in art product design with students from the School of Design and Art, University of G. The results show that the liking degree of three-color matching in the cognitive experiment test is 0.307, which is higher than the liking degree of two-color matching of 0.223, indicating that overall three-color matching samples are more popular in art design. This study provides effective evaluation and guidance for designing art products that are emotional and innovative.

Semantic-Layout-Guided Image Synthesis for High-Quality Synthetic-Aperature Radar Detection Sample Generation

With the widespread application and functional complexity of deep neural networks (DNNs), the demand for training samples is increasing. This elevated requirement also extends to DNN-based SAR object detection. Most public SAR object detection datasets are oriented to marine targets such as ships, while data sets oriented to land targets are relatively rare, though they are an effective way to improve the land object detection capability of deep models through SAR sample generation. In this paper, a synthesis generation collaborative SAR sample augmentation framework is proposed to achieve flexible and diverse high-quality sample augmentation. First, a semantic-layout-guided image synthesis strategy is proposed to generate diverse detection samples. The issues of object location rationality and object layout diversity are also addressed. Meanwhile, a pix2pixGAN network guided by layout maps is utilized to achieve diverse background augmentation. Second, a progressive training strategy of diffusion models is proposed to achieve semantically controllable SAR sample generation to further improve the diversity of scene clutter. Finally, a sample cleaning method considering distribution migration and network filtering is employed to further improve the quality of detection samples. The experimental results show that this semantic synthesis generation method can outperform existing sample augmentation methods, leading to a comprehensive improvement in the accuracy metrics of classical detection networks.

Hashing Fake: Producing Adversarial Perturbation for Online Privacy Protection Against Automatic Retrieval Models

The wide application of deep neural networks (DNNs) has significantly improved the performance of hashing models on multimodal retrieval issues. DNN-based deep models can automatically learn semantic features from raw data to make human-level decisions. However, the superior generalization leads to potential privacy leakage risks. Strong DNN-based retrieval models enable malicious crawlers to search for nontag private information based on semantic similarity matching. Hence, executing effective privacy protection mechanisms against those retrieval software is essential for reliable social website construction. In this article, we propose a retrieval task-based adversarial perturbation generation method called Hashing Fake to meet this request. Specifically, DNNs are recently found to be vulnerable to a specific set of attacks called adversarial perturbations, which denote some magnitude-restricted signals added on objective samples to misguide well-crafted DNN models, and perturbations’ magnitudes are small enough that will not induce humans’ perception. Moreover, since existing adversarial perturbation generation methods are designed for supervised tasks, Hashing Fake constructs a differential approximation substitution for perturbation production on unsupervised retrieval tasks. Through extensive experiments on several deep retrieval benchmarks, we demonstrate that well-crafted perturbations using Hashing Fake can effectively misguide objective models’ recognitions to make false predictions. The added norm-restricted perturbations on objective samples will not alter humans’ perception; hence, Hashing Fake can be applied on real-world social websites to protect subscribers’ privacy against malicious retrieval software.

Two‐stage architectural fine‐tuning for neural architecture search in efficient transfer learning

AbstractIn many deep neural network (DNN) applications, the difficulty of gathering high‐quality data in industry fields hinders the practical use of DNN. Thus, the concept of transfer learning (TL) has emerged, which leverages the pretrained knowledge of the DNN which was built based on large‐scale datasets. For this TL objective, this paper suggests two‐stage architectural fine‐tuning for reducing the costs and time while exploring the most efficient DNN model, inspired by neural architecture search (NAS). The first stage is mutation, which reduces the search costs using a priori architectural information. Moreover, the next stage is early‐stopping, which reduces NAS costs by terminating the search process in the middle of computation. The data‐intensive experimental results verify that the proposed method outperforms benchmarks.

Intracranial hemorrhage detection: how much can windowing, transfer learning and data augmentation affect a deep learning model’s performance?

The application of Deep Neural Networks to detect intracranial hemorrhage from computed tomography images has been widely used in clinical medicine. In general, building these solutions combines three types of approaches: Preprocessing, Transfer Learning, and Data Augmentation. This study’s goal was to measure the contribution of each of these approaches and thus highlight which approach has more margin of improvement. An experimental study was conducted on a public dataset containing computerized tomography images. The comparison used the stratified ten-fold cross-validation process to set confidence intervals evaluating performance measured by the area under the receiver operating characteristic curve. The ResNet-50 was the deep learning model selected. The results showed that all the approaches raise the generalization power when applied in isolation, and Data Augmentation offers the most significant gain to the baseline. The experiment also showed an opportunity to improve the detection of intracranial hemorrhage by applying new preprocessing techniques since this was the approach that showed the smallest increase in discriminatory power among the investigated approaches. The paired Wilcoxon’s Signed-Rank Test showed that not all the differences were statistically significant, with a confidence level of 95%.

Application of Deep Neural Network with Frequency Domain Filtering in the Field of Intrusion Detection

In the field of intrusion detection, existing deep learning algorithms have limited capability to effectively represent network data features, making it challenging to model the complex mapping relationship between network data and attack behavior. This limitation, in turn, impacts the detection accuracy of intrusion detection systems. To address this issue and further enhance detection accuracy, this paper proposes an algorithm called the Fourier Neural Network (FNN). The core of FNN consists of a Deep Fourier Neural Network Block (DFNNB), which is composed of a Hadamard Neural Network (HNN) and a Fourier Neural Network Layer (FNNL). In a DFNNB, the HNN is responsible for sampling the network intrusion data samples in different time domain spaces. The FNNL, on the other hand, performs a Fourier transform on the samples outputted by the HNN and maps them to the frequency domain space, followed by a filtering process. Finally, the data processed by filtering are transformed back to the time domain space for subsequent feature extraction work by the DFNNB. Additionally, to enhance the algorithm’s detection accuracy and filter out noise signals, this paper also introduces a High-energy Filtering Process (HFP), which eliminates noise signals from the data signal and reduces interference on the final detection result. Due to the ability of FNN to process network data in both the time domain space and the frequency domain space, it possesses a stronger capability in expressing data features. Finally, this paper conducts performance evaluations on the KDD Cup99, NSL-KDD, UNSW-NB15, and CICIDS2017 datasets. The results demonstrate that the proposed FNN-based IDS model achieves higher detection rates, lower false alarm rates, and better detection performance than classical deep learning and machine learning methods.