Zero-Bias Deep Learning for Accurate Identification of Internet-of-Things (IoT) Devices

Liu, Yongxin; Wang, Jian; Li, Jianqiang; Song, Houbing; Yang, Tao; Niu, Shuteng; Ming, Zhong

doi:10.1109/jiot.2020.3018677

Cited by 86 publications

(38 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…are more important for parsing together melodic structures (similarity ratings, melodic repertoire, etc.) [19][20][21][22][23].…”

Section: Related Workmentioning

confidence: 99%

[Retracted] Research on the Model of Music Sight‐Singing Guidance System Based on Artificial Intelligence

Zhe

2021

Complexity

View full text Add to dashboard Cite

Intelligent Guided Learning System (ITS) is a computer system that uses computers to imitate the experience and methods of teaching experts to assist in teaching; ITS provides learners with personalized learning resources and adaptive teaching methods, thus reducing students’ dependence on teachers and realizing independent learning. After years of research and development, and with the help of artificial intelligence technology, ITS has basically formed a stable structure and unified implementation specifications and has given birth to many excellent products in disciplines such as basic computer, language, medicine, and mathematics. Based on the above background, this paper takes the system structure of universal ITS as the basis, combines the characteristics of music sight-singing subject in teaching contents and teaching methods, researches the intelligent guidance system model of music sight-singing, and completes the design of the overall system architecture. The specific strategies and implementation methods of the teaching methods, resource recommendation, and ability assessment components of the teaching model are studied. The research includes the definition of the difficulty characteristics of the score, the design of the score recommendation algorithm, the design of the sight-singing scoring algorithm, and the experimental analysis of each algorithm. It is proposed that the difficulty feature-based score recommendation algorithm is the core component of resource recommendation in the teacher model, and the sight-singing scoring algorithm is the basis for ability assessment and learner model update.

show abstract

“…are more important for parsing together melodic structures (similarity ratings, melodic repertoire, etc.) [19][20][21][22][23].…”

Section: Related Workmentioning

confidence: 99%

[Retracted] Research on the Model of Music Sight‐Singing Guidance System Based on Artificial Intelligence

Zhe

2021

Complexity

View full text Add to dashboard Cite

show abstract

“…Where RU (•) and CU (•) denote deriving column-wise and row-wise direction vectors (vectors' magnitudes are normalized to one) of their inputs. Our prior results [32], [33] prove that the zero-bias dense layer can work seamlessly with backpropagation mechanisms and trained using regular loss functions (e.g., binary crossentropy, etc. even if L 2 can be replaced by a regular dense layer, it can also be viewed as a similarity matching layer, but the matching results are weighted and biased [32].…”

Section: Methodology a Zero-bias Deep Neural Network For Wireless Device Identificationmentioning

confidence: 70%

“…residuals, respectively. We have discovered that the last dense layer of a DNN classifier performs the nearest neighbor matching with biases and preferabilities using cosine similarity, We also show that a DNN classifier's accuracy will not be impaired if we replace its last dense layer with a zero-bias dense layer [32], in which the decision biases and preferabilities are eliminated. We can denote its mechanism as (also in Figure 2):…”

Section: Methodology a Zero-bias Deep Neural Network For Wireless Device Identificationmentioning

confidence: 85%

“…A typical way to implement smart decision functionality in IoT is by integrating learning-enabled components through Deep Learning (DL) and Deep Neural Networks (DNNs). One typical application of DNNs in IoT is the passive identification of IoT devices through their wireless signals for Non-cryptographic Device Identification (NDI) and Physical Layer authentication [8]- [10]. DL and DNNs are effective in wireless device identification under various scenarios, however, DNN models in these applications need to continuous evolving to adapt to operational variations as new devices (as new classes) are emerging.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Class-Incremental Learning for Wireless Device Identification in IoT

Liu¹,

Wang²,

Li³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

<div>Deep Learning (DL) has been utilized pervasively in the Internet of Things (IoT). One typical application of DL in IoT is device identification from wireless signals, namely Non-cryptographic Device Identification (NDI). However, learning components in NDI systems have to evolve to adapt to operational variations, such a paradigm is termed as Incremental Learning (IL). Various IL algorithms have been proposed and many of them require dedicated space to store the increasing amount of historical data, and therefore, they are not suitable for IoT or mobile applications. However, conventional IL schemes can not provide satisfying performance when historical data are not available. In this paper, we address the IL problem in NDI from a new perspective, firstly, we provide a new metric to measure the degree of topological maturity of DNN models from the degree of conflict of class-specific fingerprints. We discover that an important cause for performance degradation in IL enabled NDI is owing to the conflict of devices' fingerprints. Second, we also show that the conventional IL schemes can lead to low topological maturity of DNN models in NDI systems. Thirdly, we propose a new Channel Separation Enabled Incremental Learning (CSIL) scheme without using historical data, in which our strategy can automatically separate devices' fingerprints in different learning stages and avoid potential conflict. Finally, We evaluated the effectiveness of the proposed framework using real data from ADS-B (Automatic Dependent Surveillance-Broadcast), an application of IoT in aviation. The proposed framework has the potential to be applied to accurate identification of IoT devices in a variety of IoT applications and services. Data and code available at IEEE Dataport (DOI: 10.21227/1bxc-ke87) and \url{https://github.com/pcwhy/CSIL}}.<br></div>

show abstract

“…The decisional memory representations usually exist within the last dense layer of neural networks. And in this document, we do not consider the bias neurons and amplificative attentions, because we have proved that such a simplification will not impair the performance of neural networks [3], [4].…”

mentioning

confidence: 99%

Formal Proofs of Orthogonality for Class-Incremental Learning for Wireless Device Identification in IoT

Liu¹,

Wang²,

Li³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

<div> <div> <div> <p>This document provides a formal proof and supple- mentary information of the paper: Class-Incremental Learning for Wireless Device Identification in IoT. The original paper focuses on providing a novel and efficient incremental learning algorithm. In this document, we explicitly explain why the mem- ory representations (latent device fingerprints in our application) in Artificial Neural Networks approximate orthogonality with insights for the invention of our Channel Separation Incremental Learning algorithm. </p> </div> </div> </div>

show abstract

Zero-Bias Deep Learning for Accurate Identification of Internet-of-Things (IoT) Devices

Cited by 86 publications

References 25 publications

[Retracted] Research on the Model of Music Sight‐Singing Guidance System Based on Artificial Intelligence

[Retracted] Research on the Model of Music Sight‐Singing Guidance System Based on Artificial Intelligence

Class-Incremental Learning for Wireless Device Identification in IoT

Formal Proofs of Orthogonality for Class-Incremental Learning for Wireless Device Identification in IoT

Contact Info

Product

Resources

About