Federated Learning in Mobile Edge Networks: A Comprehensive Survey

IEEE Communications Surveys & Tutorials, Journal Year: 2020, Volume and Issue: 22(3), P. 2031 - 2063

Published: Jan. 1, 2020

In recent years, mobile devices are equipped with increasingly advanced sensing and computing capabilities. Coupled advancements in Deep Learning (DL), this opens up countless possibilities for meaningful applications, e.g., medical purposes vehicular networks. Traditional cloud-based Machine (ML) approaches require the data to be centralized a cloud server or center. However, results critical issues related unacceptable latency communication inefficiency. To end, Mobile Edge Computing (MEC) has been proposed bring intelligence closer edge, where is produced. conventional enabling technologies ML at edge networks still personal shared external parties, servers. Recently, light of stringent privacy legislations growing concerns, concept Federated (FL) introduced. FL, end use their local train an model required by server. The then send updates rather than raw aggregation. FL can serve as technology since it enables collaborative training also DL network optimization. large-scale complex network, heterogeneous varying constraints involved. This raises challenges costs, resource allocation, security implementation scale. survey, we begin introduction background fundamentals FL. Then, highlight aforementioned review existing solutions. Furthermore, present applications Finally, discuss important future research directions

Language: Английский

---

Molecular Docking: Shifting Paradigms in Drug Discovery

International Journal of Molecular Sciences, Journal Year: 2019, Volume and Issue: 20(18), P. 4331 - 4331

Published: Sept. 4, 2019

Molecular docking is an established in silico structure-based method widely used drug discovery. Docking enables the identification of novel compounds therapeutic interest, predicting ligand-target interactions at a molecular level, or delineating structure-activity relationships (SAR), without knowing priori chemical structure other target modulators. Although it was originally developed to help understanding mechanisms recognition between small and large molecules, uses applications discovery have heavily changed over last years. In this review, we describe how firstly applied assist tasks. Then, illustrate newer emergent docking, including prediction adverse effects, polypharmacology, repurposing, fishing profiling, discussing also future further potential technique when combined with techniques, such as artificial intelligence.

Language: Английский

---

Deep Learning in Mobile and Wireless Networking: A Survey

IEEE Communications Surveys & Tutorials, Journal Year: 2019, Volume and Issue: 21(3), P. 2224 - 2287

Published: Jan. 1, 2019

The rapid uptake of mobile devices and the rising popularity applications services pose unprecedented demands on wireless networking infrastructure. Upcoming 5G systems are evolving to support exploding traffic volumes, real-time extraction fine-grained analytics, agile management network resources, so as maximize user experience. Fulfilling these tasks is challenging, environments increasingly complex, heterogeneous, evolving. One potential solution resort advanced machine learning techniques, in order help manage rise data volumes algorithm-driven applications. recent success deep underpins new powerful tools that tackle problems this space. In paper, we bridge gap between research, by presenting a comprehensive survey crossovers two areas. We first briefly introduce essential background state-of-the-art techniques with networking. then discuss several platforms facilitate efficient deployment onto systems. Subsequently, provide an encyclopedic review research based learning, which categorize different domains. Drawing from our experience, how tailor environments. complete pinpointing current challenges open future directions for research.

Language: Английский

---

Review of Deep Learning Algorithms and Architectures

IEEE Access, Journal Year: 2019, Volume and Issue: 7, P. 53040 - 53065

Published: Jan. 1, 2019

Deep learning (DL) is playing an increasingly important role in our lives. It has already made a huge impact areas, such as cancer diagnosis, precision medicine, self-driving cars, predictive forecasting, and speech recognition. The painstakingly handcrafted feature extractors used traditional learning, classification, pattern recognition systems are not scalable for large-sized data sets. In many cases, depending on the problem complexity, DL can also overcome limitations of earlier shallow networks that prevented efficient training abstractions hierarchical representations multi-dimensional data. neural network (DNN) uses multiple (deep) layers units with highly optimized algorithms architectures. This paper reviews several optimization methods to improve accuracy reduce time. We delve into math behind recent deep networks. describe current shortcomings, enhancements, implementations. review covers different types architectures, convolution networks, residual recurrent reinforcement variational autoencoders, others.

Language: Английский

---

A State-of-the-Art Survey on Deep Learning Theory and Architectures

Electronics, Journal Year: 2019, Volume and Issue: 8(3), P. 292 - 292

Published: March 5, 2019

In recent years, deep learning has garnered tremendous success in a variety of application domains. This new field machine been growing rapidly and applied to most traditional domains, as well some areas that present more opportunities. Different methods have proposed based on different categories learning, including supervised, semi-supervised, un-supervised learning. Experimental results show state-of-the-art performance using when compared approaches the fields image processing, computer vision, speech recognition, translation, art, medical imaging, information robotics control, bioinformatics, natural language cybersecurity, many others. survey presents brief advances occurred area Deep Learning (DL), starting with Neural Network (DNN). The goes cover Convolutional (CNN), Recurrent (RNN), Long Short-Term Memory (LSTM) Gated Units (GRU), Auto-Encoder (AE), Belief (DBN), Generative Adversarial (GAN), Reinforcement (DRL). Additionally, we discussed developments, such advanced variant DL techniques these approaches. work considers papers published after 2012 from history began. Furthermore, explored evaluated domains are also included this survey. We recently developed frameworks, SDKs, benchmark datasets used for implementing evaluating There surveys neural networks (RL). However, those not individual training large-scale models method generative models.

Language: Английский

---

Deep Learning for IoT Big Data and Streaming Analytics: A Survey

IEEE Communications Surveys & Tutorials, Journal Year: 2018, Volume and Issue: 20(4), P. 2923 - 2960

Published: Jan. 1, 2018

In the era of Internet Things (IoT), an enormous amount sensing devices collect and/or generate various sensory data over time for a wide range fields and applications. Based on nature application, these will result in big or fast/real-time streams. Applying analytics such streams to discover new information, predict future insights, make control decisions is crucial process that makes IoT worthy paradigm businesses quality-of-life improving technology. this paper, we provide thorough overview using class advanced machine learning techniques, namely deep (DL), facilitate domain. We start by articulating characteristics identifying two major treatments from perspective, streaming analytics. also discuss why DL promising approach achieve desired types The potential emerging techniques are then discussed, its promises challenges introduced. present comprehensive background different architectures algorithms. analyze summarize reported research attempts leveraged smart have incorporated their intelligence discussed. implementation approaches fog cloud centers support applications surveyed. Finally, shed light some directions research. At end each section, highlight lessons learned based our experiments review recent literature.

Language: Английский

---

A survey on deep learning for big data

Information Fusion, Journal Year: 2017, Volume and Issue: 42, P. 146 - 157

Published: Nov. 11, 2017

Language: Английский

---

An Intelligent Fault Diagnosis Method Using Unsupervised Feature Learning Towards Mechanical Big Data

IEEE Transactions on Industrial Electronics, Journal Year: 2016, Volume and Issue: 63(5), P. 3137 - 3147

Published: Jan. 19, 2016

Intelligent fault diagnosis is a promising tool to deal with mechanical big data due its ability in rapidly and efficiently processing collected signals providing accurate results. In traditional intelligent methods, however, the features are manually extracted depending on prior knowledge diagnostic expertise. Such processes take advantage of human ingenuity but time-consuming labor-intensive. Inspired by idea unsupervised feature learning that uses artificial intelligence techniques learn from raw data, two-stage method proposed for machines. first stage method, sparse filtering, an two-layer neural network, used directly vibration signals. second stage, softmax regression employed classify health conditions based learned features. The validated motor bearing dataset locomotive dataset, respectively. results show obtains fairly high accuracies superior existing methods dataset. Because adaptively, reduces need labor makes handle more easily.

Language: Английский

---

Machine learning on big data: Opportunities and challenges

Neurocomputing, Journal Year: 2017, Volume and Issue: 237, P. 350 - 361

Published: Jan. 12, 2017

Language: Английский

---

A Survey of Machine and Deep Learning Methods for Internet of Things (IoT) Security

IEEE Communications Surveys & Tutorials, Journal Year: 2020, Volume and Issue: 22(3), P. 1646 - 1685

Published: Jan. 1, 2020

The Internet of Things (IoT) integrates billions smart devices that can communicate with one another minimal human intervention. IoT is the fastest developing fields in history computing, an estimated 50 billion by end 2020. However, crosscutting nature systems and multidisciplinary components involved deployment such have introduced new security challenges. Implementing measures, as encryption, authentication, access control, network application for their inherent vulnerabilities ineffective. Therefore, existing methods should be enhanced to effectively secure ecosystem. Machine learning deep (ML/DL) advanced considerably over last few years, machine intelligence has transitioned from laboratory novelty practical machinery several important applications. Consequently, ML/DL are transforming merely facilitating communication between security-based systems. goal this work provide a comprehensive survey ML recent advances DL used develop threats related or newly presented, various potential system attack surfaces possible each surface discussed. We then thoroughly review present opportunities, advantages shortcomings method. discuss opportunities challenges applying security. These serve future research directions.

Language: Английский

---