論文 - 大槻　知明

Reinforcement Learning With Clustering Optimization for Antenna Parameter Adjustment in HAPS Networks

Bouazizi M., Yang S., Li Z., Yin Y., Ohtsuki T.

IEEE Access 13   172062 - 172079 2025年

　概要を見る

High-Altitude Platform Stations (HAPS) promise to extend broadband coverage over wide areas by acting as quasi-stationary aerial base stations. However, wind-induced movements cause antenna misalignment and unbalanced user distributions, leading to throughput degradation. This work presents a low-complexity, equal clustering (EC) -aided deep Q-network (DQN) algorithm for dynamic antenna control in HAPS networks. The proposed EC algorithm partitions users into balanced clusters by identifying low-density sectors in a single pass, avoiding the iterative refinements of other clustering algorithms such as K-Means. A DQN then fine-tunes the antenna tilt angles and beam widths based on the EC results to maximize the throughput of low-throughput users. The proposed method’s performance is evaluated through simulations under rotation and shift scenarios and benchmarked against three Reinforcement Learning (RL)-based approaches and a K-Means clustering baseline. Compared with these methods, the EC-aided DQN delivers higher throughput, especially for users in the low-throughput tail, while reducing computational complexity. For instance, the percentage of low-throughput users was reduced by over 10% in the case of 5 km shift in the regions of Osaka and Sendai, compared to K-means. Similarly, in the case of 30° rotation, the percentage is reduced by 4.1% and 6.5% for the same regions. Statistical significance tests confirm that the improvements are not due to random simulation noise.

• Access to Document (DOI)

Nullforming Strategy Based on User Distribution for Spectrum Sharing Between High-Altitude Platforms and Terrestrial Networks

Fontaine K., Kalør A.E., Ohtsuki T., Ishikawa T.

IEEE Vehicular Technology Conference 2025年

ISSN  10903038

　概要を見る

High-Altitude Platform Stations (HAPSs) enable wide-area coverage in 6G networks but introduce interference when sharing spectrum with terrestrial networks (TNs). Null-forming is a technique to mitigate this interference by directing low-power beams (nulls) toward terrestrial users. Traditional nullforming methods, such as Null-Sweeping, rely on changing null directions across the resource blocks (RB) to improve the impact of nullforming. Yet these null directions are predefined for uniform user distributions and may not fully account for nonuniform deployments. We propose a user-aware nullforming approach that leverages K-means clustering to adapt null positions to dense user regions in the terrestrial cells, while time-frequency resources are allocated proportionally to user density. Simulations show that our method reduces HAPS interference for terrestrial users and improves fairness in interference distribution.

• Access to Document (DOI)

Non-contact Heart Rate Estimation Using VMD with Spectrogram-Based Preprocessing via CW Radar

Wang X., Bouazizi M., Ohtsuki T.

Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society EMBS 2025年

ISSN  1557170X

　概要を見る

Vital signs such as heart rate are important indicators for assessing human health. In particular, noninvasive heart rate measurement using non-contact devices such as continuous wave (CW) radar has attracted attention as a promising technique for estimating heart rate by detecting small cardiovascular movements. However, existing estimation methods have difficulty in accurately estimating heart rate when strongly affected by noise such as respiration and body movement. In this research, we propose a heart rate estimation method with high robustness against noise by combining a pre-processing method using spectrogram and signal decomposition based on Variational Mode Decomposition (VMD). In the proposed method, preprocessing based on the spectrogram is first performed before applying VMD. The obtained integrated spectrum containing the heartbeat-related components is then processed using VMD, which decomposes it into multiple intrinsic mode function (IMF) signals. To select the IMF signals containing heartbeat components, the autocorrelation function and the frequency variation characteristic of heartbeat signals are used as criteria for selection. To evaluate the characteristics of the proposed method, measurements were performed on 10 subjects in a sitting position while electrocardiogram (ECG) was simultaneously recorded as a reference signal. Experimental results showed that the proposed method achieved a low mean absolute error (MAE) of 5.46 bpm compared to the conventional methods, effectively suppressing the effect of noise and improving the accuracy of heart rate estimation.

• Access to Document (DOI)

NoisyNet-DDQN-Based Sequential Handoff Algorithm for UAV Networks With End-to-End Network Slicing

Yang F., Li M., Ohtsuki T., Wu W., Si P., Sun T.

IEEE Transactions on Vehicular Technology 74 （ 12 ） 19496 - 19512 2025年

ISSN  00189545

　概要を見る

End-to-end network slicing in uncrewed aerial vehicle (UAV) networks has attracted significant attention for delivering differentiated services tailored to the quality of service (QoS) requirements of user terminals (UTs). In a sliced UAV network, the dual mobility of both UTs and UAVs, combined with the three-layer association of UT-UAV-network slice (NS), makes handoff a critical and challenging problem for maintaining the communication performance of UTs. This paper proposes a sequential handoff process where a central controller makes handoff decisions for each UT sequentially according to the UT queue. This process is modeled as a Markov decision process (MDP). In the MDP, the state space includes NS available resources, UTs' QoS requirements, and UT queue information. Additionally, the reward function accounts for UTs' QoS and handoff overhead. The double deep Q-network (DDQN) technique is then employed to design a handoff algorithm. This algorithm achieves the optimization goal of ensuring UTs' QoS requirements while avoiding frequent handoffs by maximizing the total reward of the MDP. Additionally, a state-dependent exploration technique with adaptive intensity adjustment, called noisy network (NoisyNet), is utilized to drive the DDQN for effective exploration, thereby enhancing the performance of the handoff algorithm. Finally, the runtime complexity of the proposed NoisyNet-DDQN-based sequential handoff (NoisyNet-DDQN-SH) algorithm is analyzed. The simulation results show that the NoisyNet-DDQN-SH algorithm converges well. Moreover, it exhibits strong robustness. Additionally, the NoisyNet technique effectively enhances the algorithm's performance, making the NoisyNet-DDQN-SH algorithm superior to the vanilla DDQN (with an \varepsilon-greedy strategy), greedy, maximum signal-to-interference-plus-noise ratio (Max-SINR), and random algorithms.

• Access to Document (DOI)

Multi-Dimensional Spectrum Prediction Method Based on Efficient Adaptive Broad Learning

Ji N., Wang S., Zhang Y., Peng Y., Otsuki T., Niyato D., Gui G.

IEEE Wireless Communications and Networking Conference Wcnc 2025年

ISSN  15253511

　概要を見る

Multi-dimensional spectrum prediction is essential for spectrum sharing and dynamic spectrum access (DSA), tack-ling spectrum scarcity and improving wireless communication. Traditional methods often use machine learning (ML), which requires manual feature extraction, or deep learning (DL), which demands high computational resources. This paper proposes a lightweight multi-dimensional spectrum prediction model using an adaptive broad learning network (ABLN). The model employs a sliding window to preprocess data and establishes input layers using randomly generated feature and enhancement nodes. The weights of broad learning are determined by solving the pseudo-inverse, and the structure is incrementally extended without retraining, reducing computational complexity. An adaptive node increment module optimizes hyperparameters efficiently. Experimental results demonstrate that ABLN reduces computational overhead while maintaining robust prediction performance across various scenarios.

• Access to Document (DOI)

More is Better: Channel-Robust Radio Frequency Fingerprinting with Random Overlay Augmentation

Wang Y., Meneghello F., Wang S., Otsuki T., Yuen C., Gui G., Wang X.

IEEE Wireless Communications and Networking Conference Wcnc 2025年

ISSN  15253511

　概要を見る

Radio Frequency Fingerprinting (RFF) is a critical technology for enhancing physical-layer security by leveraging the unique RF characteristics of hardware, enabling authentication and anti-counterfeiting for wireless communication devices. In recent years, Deep Learning (DL) has been extensively applied in RFF, significantly improving identification accuracy and efficiency. However, DL- based RFF methods still encounter challenges regarding robustness, particularly in cross-channel scenarios. To address these challenges, we propose a channel-robust RFF method based on a Multi-Scale Convolutional Attention Network (MSCAN) with Random Overlay Augmentation (ROA). Specifically, MSCAN extracts and fuses features at different scales, allowing it to capture more comprehensive signal characteristics. Additionally, ROA is a combinatorial data augmentation (DA) strategy designed to simulate diverse characteristics of wireless propagation environments, thereby enhancing the adaptability and robustness of RFF in complex channel conditions. Experiments conducted on the ORACLE dataset demonstrate that our proposed method achieves over 92 % accuracy in cross-channel scenarios, outperforming the previously proposed DA strategy. The codes will be published in GitHub¹¹https://github.com/BeechburgPieStar/SDG-for-Robust-SEI

• Access to Document (DOI)

MobiLLM: Enabling On-Device Fine-Tuning of Billion-Sized LLMs via Server-Assisted Side-Tuning

Li L., Yang X., Wu W., Wang H., Ohtsuki T., Fu X., Pan M., Shen X.

IEEE Journal on Selected Topics in Signal Processing 19 （ 7 ） 1251 - 1265 2025年

ISSN  19324553

　概要を見る

On-device fine-tuning of large language models (LLMs) has attracted a lot of attention because of its tailoring personalized models while retaining user data locally on the mobile device. However, it faces significant challenges due to prohibitive memory requirements and slow training speeds. In this paper, we propose MobiLLM, a novel scheme enabling memory-efficient LLM fine-tuning on a single mobile device via server-assisted side-tuning. Particularly, MobiLLM strategically offloads backpropagation computations to an edge server while allowing the resource-constrained mobile device to retain merely a pretrained backbone model with frozen parameters during finetuning. It constructs a backpropagation bypass via parallel adapters decoupled from the backbone. During forward propagation, the device employs low bitwidth quantization for transmitting intermediate activations to the server to reduce communication overhead. The advantage of MobiLLM lies in: 1) confining training data strictly to the mobile device, and 2) eliminating on-device backpropagation while overlapping local computations with server execution. Collectively, MobiLLM ensures the data never leaves the local mobile device while significantly reducing mobile memory and computational burdens. We implement MobiLLM on several popular mobile devices, including NVIDIA Jetson Xavier NX and CPU-only laptops. Extensive experimental results demonstrate that MobiLLM can enable a resource-constrained mobile device to fine-tune billion-sized LLMs, achieving up to 4x memory reduction and 2.3x faster convergence as compared to state-of-the-art baselines.

• Access to Document (DOI)

mmPoint-Attention: A Unified Attention Framework for Human Pose and Activity Recognition From mmWave Radar Point Clouds

Shi X., Bouazizi M., Ohtsuki T.

IEEE Sensors Journal 25 （ 14 ） 26794 - 26805 2025年

ISSN  1530437X

　概要を見る

Millimeter-wave (mmWave) radar-based pose estimation and activity recognition provide robust, precise tracking of human movement, unaffected by lighting or environmental challenges. However, existing human pose estimation methods using point cloud data typically rely on single-frame inputs or apply voxelization, both of which have inherent limitations. Moreover, the conventional methods for activity recognition using point cloud often rely on multistep processes that lose contextual information and depend on pose accuracy. To address these issues, in this paper, we propose mmPoint-Attention, a unified attention and Transformer-based model that directly utilizes multiframe point cloud data for both pose estimation and activity recognition without voxelization. Leveraging temporal context and attention mechanisms, our model achieves accurate 3-D joint estimation and activity classification, with postprocessing to enhance sequence consistency. This approach enables high recognition accuracy and computational efficiency across diverse activities compared with conventional methods.

• Access to Document (DOI)

Memristor-Based Meta-Learning for Fast mmWave Beam Prediction in Non-Stationary Environments

Cao Y., Lu W., Ohtsuki T., Maghsudi S., Jiang X.Q., Tsimenidis C.C.

IEEE International Conference on Communications    1262 - 1267 2025年

ISSN  15503607

　概要を見る

Traditional machine learning techniques have achieved great success in improving data-rate performance and reducing latency in millimeter wave (mmWave) communications. However, these methods still face two key challenges: (i) their reliance on large-scale paired data for model training and tuning, which limits performance gains and makes beam predictions outdated, especially in multi-user mmWave systems with large antenna arrays, and (ii) meta-learning (ML)-based beamforming solutions are prone to overfitting when trained on a limited number of tasks. To address these issues, we propose a memristorbased meta-learning (M-ML) framework for predicting mmWave beam in real time. The M-ML framework generates optimal initialization parameters during the training phase, providing a strong starting point for adapting to unknown environments during the testing phase. By leveraging memory to store key data, M-ML ensures the predicted beamforming vectors are wellsuited to episodically dynamic channel distributions, even when testing and training environments do not align. Simulation results show that our approach delivers high prediction accuracy in new environments, without relying on large datasets. Moreover, MML enhances the model's generalization ability and adaptability.

• Access to Document (DOI)

Max-Min Strategy for Handover in LEO Based Non-Terrestrial Networks

Nagase R., Yang S., Ohtsuki T.

IEEE Vehicular Technology Conference 2025年

ISSN  15502252

　概要を見る

In Low-Earth Orbit (LEO)-based Non-Terrestrial Networks (NTN), stable and continuous connectivity through efficient handover (HO) mechanisms should be ensured, especially given the high mobility and dynamic satellite coverage in these networks. Conventional HO optimization approaches, such as the Service Continuity Dynamic Programming (SCDP) strategy, optimize HO but may leave some User Equipments (UEs) experiencing low link rates, reduced service time, or frequent handovers. To address this, we propose the Service Continuity Max-Min (SCMM) strategy, which employs the MaxMin algorithm to reduce the number of UEs that have poor connectivity experiences, using a uniquely defined reward function that incorporates link rate, service availability time, and HO frequency. The simulation results demonstrate that SCMM improves the minimum reward across UEs while maintaining an overall performance comparable to that of the existing SCDP methods. This approach offers a robust solution to improve service continuity and user satisfaction in NTN, making it an effective strategy for next-generation satellite networks.

• Access to Document (DOI)

Malware Traffic Classification via Expandable Class Incremental Learning With Architecture Search

Zhang X., Wang Y., Ohtsuki T., Gui G., Yuen C., Di Renzo M., Sari H.

IEEE Transactions on Information Forensics and Security 20   6074 - 6085 2025年

ISSN  15566013

　概要を見る

Malware traffic classification (MTC) is a crucial step in network intrusion detection, which is significant for network security and management. With the continuous evolution of malware traffic, traditional MTC methods are difficult to adapt efficiently to new traffic categories, and manually designed neural network structures suffer from performance bottlenecks and low design efficiency. Hence, we propose an enhanced MTC method based on expandable class incremental learning (CIL) with architecture search. The architecture search can automatically design the optimal neural network structure tailored to different network traffic characteristics, avoiding the limitations of manually designing network structures and improving classification performance. Meanwhile, expandable CIL allows the MTC model to gradually learn new traffic categories without forgetting previous knowledge, avoiding the computational overhead and efficiency loss caused by frequent retraining of the model. The experimental results demonstrate that the proposed CIL-MTC approach surpasses advanced incremental learning methods on both the Edge-IIoTset and ISCX VPN-nonVPN datasets, achieving superior classification performance while maintaining lower average trainable parameters and training costs. Especially, it achieves an average incremental accuracy of 98.55% and 99.09% on the Edge-IIoTset dataset with incremental tasks of 5 and 2, respectively.

• Access to Document (DOI)

Lightweight CSI-Based Human Activity Recognition for Multitask IoT Applications

Liu C., Miao F., Chen Y., Wu J., Wan H., Tang T., Ohtsuki T., Gui G., Sari H.

IEEE Internet of Things Journal 12 （ 17 ） 35817 - 35827 2025年

　概要を見る

As the global population continues to age and technologies, such as the Internet of Things (IoT) and edge computing advance rapidly, indoor human activity recognition (HAR)-based on Wi-Fi channel state information (CSI) has gained significant research attention. However, the high computational complexity of existing HAR methods limits their deployment on resource-constrained devices. To address this challenge, we propose a lightweight HAR method using branch decision lightweight two-stream convolution-augmented transformer (BLTHAT) model, which integrates depthwise separable convolutions (DSC) and an improved framework structure to enhance computational efficiency. Additionally, we introduce the branch fusion network (BFN), a decision-making module designed to optimize feature processing and improve model robustness. Further enhancements in attention mechanisms and regularization strategies contribute to reducing complexity while maintaining high recognition accuracy. Comprehensive experiments were conducted on a multilabel dataset. The results demonstrate that our proposed HAR method achieves high computational efficiency with minimal complexity, making it well-suited for IoT applications. Ablation studies further confirm that the multibranch structure of the BFN module enhances feature extraction without significantly increasing computational overhead.

• Access to Document (DOI)

Large Model Empowered Task Offloading for Multi- Access Edge Computing in the Internet of Vehicles

Zhou X., Guan X., Wang N., Chen H., Ohtsuki T., Zhang Y., Han Z.

IEEE Transactions on Vehicular Technology 74 （ 11 ） 18012 - 18025 2025年

ISSN  00189545

　概要を見る

The rapid advancement of the Internet of Vehicles (IoV) has created significant challenges in large-scale data processing and low-latency decision-making. Multi-access edge computing substantially enhances data processing capabilities and system response times by pushing computational resources to the network edge. Task offloading technology transfers vehicle computing tasks to edge servers, alleviating the computational burden on user devices. Traditional offloading methods struggle with the challenges posed by the rapid growth of IoV. With the advancement of deep learning, large models have emerged as powerful tools for handling complex tasks. In this paper, we propose a task offloading algorithm that integrates a Transformer-based large model with deep reinforcement learning. We utilize the Transformer-based large model to replace the action generation network in deep reinforcement learning. The proposed method utilizes extensive historical data and real-time task information to dynamically adjust task allocation schemes. This approach enables the conclusion of task offloading strategies that minimize energy consumption and latency. Additionally, we propose a task classification framework to further reduce offloading delays and energy consumption. The framework categorizes tasks based on key attributes such as computational complexity, data size, and time sensitivity. We propose a reward function based on task adaptability to conclude the task offloading strategy that minimizes both offloading delay and energy consumption. Experimental results demonstrate that the proposed algorithm can accurately and quickly conclude the task offloading strategy to minimize energy consumption and latency.

• Access to Document (DOI)

Introduction to Activities of AI and Digital Twin WG in XGMF 6G Radio Technology Project

Yamamoto T., Yamazaki T., Otsuki T., Suyama S., Liu L., Osawa N., Yamamoto H., Watanabe T., Kuriyama K., Koshimizu T., Tsukamoto Y., Tomeba H., Nagao T., Chang Y., Morita K., Kitano H.

2025 IEEE Vts Asia Pacific Wireless Communications Symposium Apwcs 2025 2025年

　概要を見る

This paper introduces activities of AI and Digital Twin working group (AI-DT WG) in XGMF 6G Radio Technology Project. Especially in the fiscal year 2024, AI-DT WG has been engaged in activities for getting the beyond 5G white paper "AI/ML and Digital Twin Technologies"ready for the world. The AI-DT WG also launched the discussion for establishing a shared dataset for AI training. This paper provides an overview of the white paper and the status of the discussion for shared dataset establishment.

• Access to Document (DOI)

GA-CBLN: A Generative Augmented Cascade Broad Learning Network for Efficient Spectrum Prediction with Missing Data Handling

Ji N., Feng Z., Wang S., Zhang Y., Otsuki T., Gui G., Sari H.

2025 IEEE International Symposium on Dynamic Spectrum Access Networks Dyspan 2025 2025年

　概要を見る

With the advancements in wireless communication and the growth of Internet of Things (IoT), spectrum scarcity has become a key bottleneck. Spectrum prediction enhances resource utilization and supports dynamic spectrum access (DSA). However, current models struggle with missing data and prioritize accuracy over efficiency, limiting practical use. To address this, we propose a generative augmented cascade broad learning network (GA-CBLN) for spectrum prediction, combining a deep learning (DL) pre-trained model with an online broad learning (BL) process. First, a DL-based regressor fills missing values by learning spatiotemporal patterns. Next, a generative adversarial network (GAN) generates synthetic samples to enrich the training set, improving adaptability. Finally, the cascade BL network (CBLN) uses the combined data for efficient prediction. Simulations on real datasets show that GA-CBLN significantly outperforms traditional methods in accuracy and efficiency, even with high missing data rates.

• Access to Document (DOI)

Enhancing Dementia Detection: Augmenting the Face Mesh

Zheng C., Bouazizi M., Ohtsuki T., Kitazawa M., Horigome T., Kishimoto T.

Proceedings International Symposium on Biomedical Imaging 2025年

ISSN  19457928

　概要を見る

Alzheimer's disease (AD) stands as a prevalent form of dementia, primarily impacting the elderly, and presently lacking any known cure. Early screening and intervention have demonstrated efficacy for both current and prospective treatments. With the global population aging, early detection of AD has become imperative. Language-based approaches for dementia detection have received considerable attention in research due to dementia's nature of hurting cognitive and language functions. Facial features, on the other hand, have been comparatively neglected despite their potential in identifying AD, due to a lack of usable dataset. AD patients show different patterns while displaying facial expressions because of Apathy, a phenomenon common in AD patients. This makes facial features non-negligible in dementia detection. The face mesh, comprising 478 landmarks, serves as a rich and informative facial representation. This study concentrates on leveraging face mesh data for AD detection. We explored data augmentation techniques to address the challenges of over-fitting, particularly impactful in training neural networks on a small dataset. Deep learning methods, including transformer, Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) networks, were employed to classify dementia patients and healthy subjects. By augmenting the dataset by rotation, shift, and scaling, we improved the performance of the deep learning methods. The combined use of 1D CNN networks and data augmentation method reported the highest accuracy of 76%. These results confirmed the effectiveness of our data augmentation methods. They also underscore the potential of face-related features in enhancing early detection approaches for AD.

• Access to Document (DOI)

Enhancing Cross-Domain Robustness in Wi-Fi-Based Human Activity Recognition via Attention-Driven Deep Learning

Miao F., Takyu O., Shan L., Ohtsuki T., Gui G.

IEEE Vehicular Technology Conference 2025年

ISSN  10903038

　概要を見る

Wi-Fi-based Human Activity Recognition (HAR) using Channel State Information (CSI) has received significant attention due to its non-intrusive nature and wide applicability in smart cities, healthcare, and smart home systems. Despite its potential, real-world deployment faces a major challenge: ensuring the model's ability to generalize across different users, who exhibit distinct behaviors and encounter diverse environmental factors. This issue, termed Cross-User Generalization (CUG), arises from distribution shifts caused by variations in users' body shapes, postures, movements, and environmental conditions. To address this challenge, we propose a cross-domain generalization framework based on an Attention-based Bidirectional Long Short-Term Memory (ABLSTM) network. The proposed framework enhances the generalization capabilities of HAR models by leveraging multi-source training and incorporating attention mechanisms for effective temporal feature learning. Specifically, the model learns user-invariant features across diverse training users and adapts to new users without requiring direct exposure to their data. Extensive experiments on a multi-user CSI dataset demonstrate that the ABLSTM model improves average test accuracy by 2.0%-6.7% compared to BiLSTM and GRU models, and over 18.0% compared to MLP-based models, while achieving the lowest training and testing loss. This methodology offers a robust and scalable solution for Wi-Fi-based HAR, promoting reliable deployment in complex, multi-user environments encountered in smart cities, healthcare monitoring, and security surveillance.

• Access to Document (DOI)

Energy Efficient Transmission Parameters Selection Method Using Reinforcement Learning in Distributed LoRa Networks

Airiyoshi R., Hasegawa M., Ohtsuki T., Li A.

IEEE Wireless Communications and Networking Conference Wcnc 2025年

ISSN  15253511

　概要を見る

With the increase in demand for Internet of Things (IoT) applications, the number of IoT devices has drastically grown, making spectrum resources seriously insufficient. Transmission collisions and retransmissions increase power consumption. Therefore, even in long-range (LoRa) networks, selecting appropriate transmission parameters, such as channel and transmission power, is essential to improve energy efficiency. However, due to the limited computational ability and memory, traditional transmission parameter selection methods for LoRa networks are challenging to implement on LoRa devices. To solve this problem, a distributed reinforcement learning-based channel and transmission power selection method is proposed, which can be implemented on the LoRa devices to improve energy efficiency in this paper. Specifically, the channel and transmission power selection problem in LoRa networks is first mapped to the multi-armed-bandit (MAB) problem. Then, an MAB-based method is introduced to solve the formulated transmission parameter selection problem based on the acknowledgment (ACK) packet and the power consumption for data transmission of the LoRa device. The performance of the proposed method is evaluated by the constructed actual LoRa network. Experimental results show that the proposed method performs better than fixed assignment, adaptive data rate low-complexity (ADR-Lite), and e-greedy-based methods in terms of both transmission success rate and energy efficiency.

• Access to Document (DOI)

Efficient WiFi Device Recognition via Blueprint Separable Residual Network with SE Module

Liu T., Cai Z., Wang Q., Ohtsuki T., Sari H., Gui G.

IEEE Vehicular Technology Conference 2025年

ISSN  10903038

　概要を見る

With the rapid advancement of wireless communication technologies, WiFi signals have become essential for modern connectivity across diverse applications. However, their widespread deployment introduces significant security vulnerabilities, including unauthorized access, data leakage, and interference. Accurate identification of WiFi transmitters is crucial for mitigating these threats. While existing methods perform well in ideal conditions, their effectiveness degrades in real-world scenarios, particularly in environments with low signal-to-noise ratios (SNRs). To address this limitation, we propose a novel transmitter identification framework that integrates blueprint separable convolution (BSC) and a squeeze-and-excitation (SE) module. The BSC extracts critical features efficiently, while the SE module dynamically enhances feature representations. Simulation results demonstrate that the proposed approach achieves competitive or superior accuracy compared to state-of-the-art models. Moreover, the framework exhibits strong robustness, maintaining high recognition performance even in challenging transmission conditions with low SNRs.

• Access to Document (DOI)

Efficient and Effective Fine-tuning Method for Depression Detection from Conversation

Ikeuchi K., Kishimoto T., Nakai F., Horigome T., Kitazawa M., Ohtsuki T.

Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society EMBS 2025年

ISSN  1557170X

　概要を見る

Depression has a significant impact on society. Early detection of depression is essential because timely treatment can mitigate its negative impact. Recent studies have reported that models based on Large Language Models (LLMs), such as Chat GPT-4, can achieve high accuracy in detecting depression when prompt-engineering techniques are applied. In contrast, fine-tuning methods require more memory for training and have yet to demonstrate comparable accuracy to prompting-engineering in detecting depression from conversations. The objective of this study is to achieve higher accuracy than prompt-engineered models with low GPU memory through fine-tuning. To accomplish this, we first quantized the parameters of the LLM to minimize memory requirements during training. Instead of adding a classification layer, we retained the original structure of the LLM and trained it as a generative model. Our evaluation on the DAIC-WOZ dataset demonstrated an F1 score of 84% for depression detection with low GPU memory, which is superior to the performance of prompt-engineered models. Additionally, when tested on the private PROMPT dataset, the proposed model achieved an F1 score of 82%. This result suggests that the model's strong performance is not solely dependent on publicly available data that may have been included in the pre-training.

• Access to Document (DOI)