Yu Gu’s research while affiliated with University of Science and Technology of China and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (59)


Conditional Convolution Residual Network for Efficient Super-Resolution
  • Chapter

September 2023

·

14 Reads

Lecture Notes in Computer Science

Yunsheng Guo

·

Jinyang Huang

·

·

[...]

·

Yu Gu

With the continuous development of deep learning, single-image super-resolution (SISR) based on convolutional neural networks (CNNs) has made significant progress. Although CNN-based methods have achieved great success, these methods are difficult to apply to edge devices due to the need for large amounts of computing resources. To address this problem, the latest advancements in efficient SISR techniques focus on reducing the number of parameters and multiply-add operations (MAdds). In this paper, we propose a novel Conditional Convolution Residual Network (CCRN) to tackle this challenge. The main idea is to use conditional convolution instead of ordinary convolutional layers for residual feature learning and to combine Contrast-aware Channel Attention (CCA) and Enhanced Spatial Attention (ESA) mechanisms to improve the model’s performance. The model’s performance is ensured while reducing the computational complexity. Experimental results demonstrate that CCRN has fewer MAdds than existing SISR methods while achieving state-of-the-art performance.


Dynamic Memory-Based Continual Learning with Generating and Screening

September 2023

·

11 Reads

Lecture Notes in Computer Science

Deep neural networks suffer from catastrophic forgetting when continually learning new tasks. Although simply replaying all previous data alleviates the problem, it requires large memory and even worse, often infeasible in real-world applications where access to past data is limited. Therefore, We propose a two-stage framework that dynamically reproduces data features of previous tasks to reduce catastrophic forgetting. Specifically, at each task step, we use a new memory module to learn the data distribution of the new task and reproduce pseudo-data from previous memory modules to learn together. This enables us to integrate new visual concepts with retaining learned knowledge to achieve a better stability-malleability balance. We introduce an N-step model fusion strategy to accelerate the memorization process of the memory module and a screening strategy to control the quantity and quality of generated data, reducing distribution differences. We experimented on CIFAR-100, MNIST, and SVHN datasets to demonstrate the effectiveness of our method.



PhyFinAtt: An Undetectable Attack Framework Against PHY Layer Fingerprint-based WiFi Authentication

January 2023

·

16 Reads

·

17 Citations

IEEE Transactions on Mobile Computing

WiFi connection has been suffering from MAC forgery attacks due to the loose authentication mechanism between access points (APs) and clients. To address this problem, the physical (PHY) layer information-based fingerprint has been adopted for safe WiFi authentication. Since such a fingerprint is constant and unique for each specific network interface card (NIC), it can effectively prevent MAC forgery attacks. However, the PHY layer information-based fingerprint is still vulnerable to malicious attacks as it is extracted from Channel State Information (CSI), and its stability can be affected by the wireless environment. In this paper, we propose a novel undetectable attack framework, called PhyFinAtt, base on which the attacker can undermine the stability of the PHY layer-based authentication fingerprints through human movement and further attack the WiFi authentication protocols. Specifically, we first demonstrate that human movement at a designated location can affect the PHY fingerprint. We then illustrate the impact of human movement on the PHY fingerprint and the relationship between the movement and the channel quality to ensure that the PHY fingerprint is destroyed by the movement in an undetected way without affecting normal communication. Extensive experiments in real-world scenarios show that our proposed attack can effectively disrupt the stability of the PHY fingerprints and significantly degrade the performance of the authentication protocols based on such fingerprints. To the best of our knowledge, this is the first study on effective attacks against the PHY information-based WiFi authentication protocols. Furthermore, we also present a practical defense mechanism without involving any additional equipment to mitigate attacks similar to PhyFinAtt.


WiGRUNT: WiFi-Enabled Gesture Recognition Using Dual-Attention Network

August 2022

·

48 Reads

·

68 Citations

IEEE Transactions on Human-Machine Systems

Gestures constitute an important form of nonverbal communication where bodily actions are used for delivering messages alone or in parallel with spoken words. Recently, there exists an emerging trend of WiFi sensing-enabled gesture recognition due to its inherent merits like remote sensing, non-line-of-sight covering, and privacy-friendly. However, current WiFi-based approaches mainly reply on domain-specific training since they don’t know “where to look” and “when to look.” To this end, we propose WiGRUNT, a WiFi-enabled gesture recognition system using dual-attention network, to mimic how a keen human being intercepting a gesture regardless of the environment variations. The key insight is to train the network to dynamically focus on the domain-independent features of a gesture on the WiFi channel state information via a spatial-temporal dual-attention mechanism. WiGRUNT roots in a deep residual network (ResNet) backbone to evaluate the importance of spatial-temporal clues and exploit their inbuilt sequential correlations for fine-grained gesture recognition. We evaluate WiGRUNT on the open Widar3 dataset and show that it significantly outperforms its state-of-the-art rivals by achieving the best-ever performance in-domain or cross-domain.



Design and Implementation of a Novel Interconnection Architecture from WiFi to ZigBee
  • Chapter
  • Full-text available

July 2022

·

27 Reads

Lecture Notes in Electrical Engineering

The signal layer heterogeneous communication technology is a cross-technology communication (CTC) technology, which is a direct communication technology between different wireless devices. Since ZigBee and WiFi have overlapping spectrum distribution, the ZigBee transmission will affect the CSI sequence. We propose a CTC technology based on machine learning and neural network, from Zigbee to WiFi, leveraging only WiFi channel state information (CSI). By classifying WiFi CSI, we can distinguish whether there is ZigBee signal transmission in WiFi signal. This paper uses the machine learning method and neural network method to classify CSI sequence analyzes the importance of CSI sequence features to the classifier, improves the accuracy of machine learning classifier by extracting multiple CSI sequence features, and improves the classification accuracy by neural network classifier. In our experimental data set, the highest accuracy can reach 95%. The evaluation results show that our accuracy is higher than the existing methods.

Download


Quantitative Measurement of Perceptual Attributes and Artifacts for Tone-Mapped HDR Display

January 2022

·

30 Reads

·

10 Citations

IEEE Transactions on Instrumentation and Measurement

Measuring electronic display quality, as perceived by human observers, has attracted high attention in current consumer displays. With limited dynamic range of consumer-level standard dynamic range (SDR) displays, high dynamic range (HDR) scenes are often rendered by different tone mapping operators (TMOs). Due to the lack of quantitative measurement and analysis of tone mapping in the existing image quality measurement (IQM) methods, it is of considerable significance to establish new IQM protocols that can differentiate the display quality of SDR electronic devices. We first propose an IQM model to exhibit the essential perceptual attributes and artifacts that are peculiar to tone mapping. Furthermore, we characterize the overall image quality (OiQ) resulting from linear regression and various machine learning techniques. Finally, the execution of without HDR reference ablation experiments demonstrates the relative contribution of these attribute measurements to OiQ. The use of IQM protocols helps with well-founded quality measurement between TMOs during tone mapping processing. Our effort is not only useful to get into the tone mapping field or when implementing a TMO but also sets the stage for quantitative measurement of TMOs. By monitoring these attributes and artifacts after different tone mapping process, user-driven or optimal display is made possible.



Citations (42)


... Therefore, FM-Fi sets itself apart from prior RF-HAR solutions by not limiting itself to HAR, because it inherits the broad recognition capability of FM. In fact, we would expect FM-Fi to be able to support other sensing tasks [12,30,32,70,71,73,76] including gesture detection [37,38,69], gait recognition [7,59,65], and even vibration monitoring [1,13,14,67,68], by modifying the target of interest; we plan to explore FM-Fi's potential beyond HAR in future work. Currently, it is still an open question if one can claim open-set capability for FM-enabled HAR [52]. ...

Reference:

Large Model for Small Data: Foundation Model for Cross-Modal RF Human Activity Recognition
PhyFinAtt: An Undetectable Attack Framework Against PHY Layer Fingerprint-based WiFi Authentication
  • Citing Article
  • January 2023

IEEE Transactions on Mobile Computing

... WiFi (IEEE 802.11) localisation is a highly researched area due to the mass availability in consumer devices and existing infrastructure [58]. RSSI, CSI (exclusive to WiFi [76]), ToF and AoA can be used with WiFi signals to achieve device localisation using WiFi access points [12], with most studies using RSSI and fingerprinting [77]. The positioning of WiFi access points is dependent on the availability of mains sockets as they require constant power [58]. ...

HED: Handling environmental dynamics in indoor WiFi fingerprint localization
  • Citing Conference Paper
  • April 2016

... The attainment of environmental effects holds significance within wireless sensing, constituting a pivotal aspect that enhances the applicability and effectiveness of sensing technologies [27]. The attribute is particularly crucial in scenarios where the deployment of sensors is dynamic or subject to frequent changes in the environment. ...

WiMate: Location-independent Material Identification Based on Commercial WiFi Devices
  • Citing Conference Paper
  • December 2021

... While these methods have achieved promising performances, they rely on the design of handcrafted features and have great limitations. The success of convolutional neural network (CNN) has given researchers a new way to solve the IQA problem [14][15][16][17][18][19]. Due to tone mapping (TM) IQA datasets [20,21] being relatively small, some CNN-based models [22] first use transfer learning strategy to training network. ...

Quantitative Measurement of Perceptual Attributes and Artifacts for Tone-Mapped HDR Display
  • Citing Article
  • January 2022

IEEE Transactions on Instrumentation and Measurement

... To verify the effectiveness of our POI in dealing with annotation confusion, we perform quantitative evaluations on the synthetic label noise of RAF-DB, FERPlus, and AffectNet. Specifically, we randomly select 10%, 20%, and 30% of the training data to flip to other emotional categories and compare them with the existing methods that resolve annotation ambiguity, such as SCN [19], DMUE [21], LRN [54], and RUL [12]. For a fair comparison, we utilize the same ResNet-18 used in the benchmark method as the backbone. ...

Mitigating Label-Noise for Facial Expression Recognition in the Wild
  • Citing Conference Paper
  • July 2022

... This approach allows for anonymization of user identities and sensitive data during processing, which not only lowers deployment costs but also enhances user comfort and privacy protection. WiFi sensing applications have grown significantly, ranging from simple object detection [5] to more complex tasks such as motion recognition [6], gesture recognition [7], indoor positioning [8], and even monitoring vital signs like respiratory rate and heart rate [9]. ...

WiGRUNT: WiFi-Enabled Gesture Recognition Using Dual-Attention Network
  • Citing Article
  • August 2022

IEEE Transactions on Human-Machine Systems

... For example, ref. [16] proposes an innovative model called space-temporal convolution with nested long short-term memory (STC-NLSTMNet), a novel architecture that extracts spatial and temporal features by combining spatio-temporal convolution with nested long short-term memory network (NLSTM). And another study [17] combines the channel attention mechanism with the CNN-LSTM model, which dynamically adjusts the channel feature weights on the basis of extracting temporal and spatial features and assigns higher weights to the key channels. These methods have achieved better results in integrating local features and time dependence, but are still insufficient in capturing long-distance dependence and global features. ...

A Novel WiFi Gesture Recognition Method Based on CNN-LSTM and Channel Attention
  • Citing Conference Paper
  • November 2021

... Wital is a real-time vital signs monitoring system based on low-cost and widely available off-the-shelf Wi-Fi devices [21]. The detection signal arises from the deformations of the abdomen and chest caused by breathing and heartbeat, and these deformations can affect the propagation of Wi-Fi signals, recorded by the Wi-Fi Chanel State Information. ...

Real-time Vital Signs Monitoring Based on COTS WiFi Devices
  • Citing Conference Paper
  • December 2021

... This involves combining real-time data captured by multiple sensors with information available in related databases to obtain more accurate data. Currently, data fusion technology finds applications across various fields, where it has proven to be highly beneficial, such as face recognition [5][6], disease diagnosis [7], target tracking [8], etc. In real life, many data are time series data. ...

3-D Facial Expression Recognition via Attention-Based Multichannel Data Fusion Network
  • Citing Article
  • November 2021

IEEE Transactions on Instrumentation and Measurement