Currently, large language models (LLMs) have made significant progress in the field of psychological counseling.However, existing mental health LLMs overlook a critical issue where they do not consider the fact that different psychological counselors exhibit different personal styles, including linguistic styles and therapeutic types, etc.As a result, these LLMs fail to satisfy the individual needs of clients who seek different counseling styles.To help bridge this gap, we propose PsyDT, a novel framework using LLMs to construct the Digital Twin of Psychological counselor with personalized counseling style.Compared to the timeconsuming and costly approach of collecting a large number of real-world counseling cases to create a specific counselor's digital twin, our framework offers a faster and more costeffective solution.To construct PsyDT, we utilize dynamic one-shot learning by using GPT-4 to capture counselor's unique counseling style, mainly focusing on linguistic style and therapy technique.Subsequently, using existing singleturn long-text dialogues with client personality, GPT-4 is guided to synthesize multi-turn dialogues of specific counselor.Finally, we finetune the LLMs on the synthesized dataset, Psy-DTCorpus, to achieve the digital twin of psychological counselor with personalized counseling style.Experimental results indicate that our proposed PsyDT framework can synthesize multi-turn dialogues that closely resemble realworld counseling cases and demonstrate better performance compared to other baselines, thereby show that our framework can effectively construct the digital twin of psychological counselor with a specific counseling style. 1

In this paper, we propose LoRaSeek, a lightweight and reliable LoRa denoising framework that enhances signal quality and robustness for neural-enhanced LoRa decoding. LoRaSeek integrates a hybrid architecture combining Convolutional Neural Networks (CNNs), Transformers, and a hierarchical U-Net to effectively capture multi-scale, multidimensional features of LoRa chirp signals. To maintain efficiency, we integrate a lightweight Transformer block that supports various LoRa configurations while keeping computational overhead low. Additionally, we incorporate dual attention-based skip connections to preserve chirp signal properties across different scales. Experiments across diverse LoRa configurations show that LoRaSeek achieves 2.04–3.86 dB signal-to-noise ratio (SNR) gains over standard decoding methods and up to 3.03 dB improvement over state-of-the-art neural-enhanced LoRa decoding methods while reducing model storage by up to 7.4× and inference time by up to 1.6×.

Accurate leaf wetness detection is essential to understanding plant health and growth conditions. The mmWave radar, with its sensitivity to subtle changes, is well-suited for leaf wetness detection. Existing mmWave-based approaches utilize the Synthetic Aperture Radar (SAR) algorithm to generate image-like inputs and rely on multi-modality fusion with an RGB camera to classify leaf wetness. However, the lack of understanding of SAR-based mmWave imaging limits its accuracy in various environments. This paper presents Proteus, a novel way of understanding mmWave SAR imaging. We design a noise reduction algorithm to reduce speckle noise and improve image clarity for SAR-based mmWave imaging. Then, we incorporate phase angle data to enrich SAR texture information to capture high-resolution surface details, increasing informative features for precise wetness assessment in complex plant structures. Additionally, we introduce a cross-modality Teacher-Student network, using an RGB-based teacher model to guide the mmWave SAR-based student model for feature extraction. This network transfers the explicit knowledge in the RGB image domain to the mmWave image domain. We use commercial-off-the-shelf mmWave radar to prototype Proteus. The evaluation results show that Proteus achieves up to 96.3% accuracy across varied environmental scenarios, outperforming state-of-the-art methods.

LoRa, as a typical representative of Low Power Wide Area Networks (LPWAN), has been widely used to connect massive IoT devices. However, in mobile applications, there is massive packet loss in LoRa transmission due to link performance degradation, especially when LoRa end-devices move far from the gateway or into obstructed areas. Existing studies take little account of end-device movement, particularly when the movement pattern is unknown. We propose LMLoRa to enhance the Link Performance for Mobile LoRa networks in general scenarios. The key observation is that repeating the original packet content enhances link performance and allows smaller and more energy-efficient transmission parameter selections. Technically, LMLoRa proposes a link performance estimation model for mobile LoRa networks based on packet content repetition. Second, we exploit key hardware features of LoRa to obtain much continuous RSSI information for link quality prediction. Additionally, LMLoRa develops a channel frequency allocation policy to mitigate transmission collisions. Finally, LMLoRa designs a communication mechanism to assist the estimation model and work the whole system with low communication overhead. We design and implement LMLoRa in complex realworld environments, results show that LMLoRa enhances packet reception rate by 33.4% and energy efficiency by 14.4% on average compared with the state-of-the-art.

LoRa Localization, a fundamental service in LoRa networks, has garnered significant attention due to its long-range capabilities and low power consumption. However, existing approaches for LoRa localization are either incompatible with commercial devices or highly susceptible to environmental factors. To tackle this challenge, we propose SyncLoc, a TDoA-based LoRa localization framework that integrates a dedicated node for multi-dimensional time-drift correction. Our proposal is built on two key observations: firstly, the nanosecond-level measurement of time differences between gateways, and secondly, the substantial impact of SNR on gateway time drift. To accomplish our objective, we present three progressively enhanced versions of SyncLoc, each intended to comprehensively analyze the factors influencing LoRa time synchronization accuracy across different deployment scenarios involving nodes, carrier frequencies, and spreading factors. In addition to improving accuracy, we identify inefficiencies in LoRa’s multi-node concurrent localization, and introduce SyncLoc-4, a multi-node localization scheduling mechanism that optimizes efficiency with a 2-approximation ratio. Extensive experiments utilizing commercial LoRa devices in real-world demonstrates a 2.44× improvement in accuracy. Furthermore, simulations of large-scale networks exhibit a 2.47× boost in localization scalability (i.e., the number of concurrently located nodes) when employing SyncLoc instead of LoRaWAN.

As an emerging Low-Power Wide Area Networks (LPWAN) technology, LoRa is dedicated to providing long-range connections for pervasive Internet-of-Things devices. As LoRa operates in the unlicensed spectrum with an ALOHA-based MAC-layer protocol stack, transmissions from multiple LoRa end-devices inevitably collide with each other, leading to packet losses and increased transmission delay. Targeting at collisions caused by interferences under thesamespreading factor (SF) settings, researchers introduce multiple lines of techniques. Despite their efforts, these techniques commonly neglect the potential collisions caused by interferences underdifferentSF settings, resulting in imperfect orthogonality. Given the disparate transmission power configurations and diverse deployed locations, the collisions under different SFs commonly exist in practical networks and significantly limit the LoRa reliability. This paper presents X-MAC, the first scheduler aware of imperfect orthogonality. Technically, X-MAC detects the collisions under different SFs via tracking historical transmissions, and performs dynamic channel scheduling to avoid collisions caused by interferences under the same and different SFs. Extensive evaluations on testbed devices show that, compared with the state-of-the-art methods, X-MAC boosts the network scalability (number of concurrent end-devices) by 1.26× to 2.41× with packet reception rate requirement of > 95%.

Facial image-based kinship verification is a rapidly growing field in computer vision and biometrics. The key to determining whether a pair of facial images has a kin relation is to train a model that can enlarge the margin between the faces that have no kin relation while reducing the distance between faces that have a kin relation. Most existing approaches primarily exploit duplet (i.e., two input samples without cross pair) or triplet (i.e., single negative pair for each positive pair with low-order cross pair) information, omitting discriminative features from multiple negative pairs. These approaches suffer from weak generalizability, resulting in unsatisfactory performance. Inspired by human visual systems that incorporate both low-order and high-order cross-pair information from local and global perspectives, we propose to leverage high-order cross-pair features and develop a novel end-to-end deep learning model called the adaptively weighted k -tuple metric network (AW k -TMN). Our main contributions are three-fold. First, a novel cross-pair metric learning loss based on k -tuplet loss is introduced. It naturally captures both the low-order and high-order discriminative features from multiple negative pairs. Second, an adaptively weighted scheme is formulated to better highlight hard negative examples among multiple negative pairs, leading to enhanced performance. Third, the model utilizes multiple levels of convolutional features and jointly optimizes feature and metric learning to further exploit the low-order and high-order representational power. Extensive experimental results on three popular kinship verification datasets demonstrate the effectiveness of our proposed AW k -TMN approach compared with several state-of-the-art approaches. The source codes and models are released.1.

Wireless sensing has emerged as a powerful environmental sensing technology that is vulnerable to the impact of all kinds of ambient noises. LoRa is a novel interference-resilient technology of low-power wide-area networks (LPWAN), which has attracted wide attention from scientific and industrial communities. However, LoRa transmission suffers from serious latency in those complex wireless sensing environments requiring transmission reliability. In this article, we present CH-MAC, the first MAC-layer protocol based on the local corruption nature of packets and the time-varying nature of channels to reduce end-to-end transmission latency in LPWAN with reliable communication requirements. Specifically, CH-MAC employs Luby Transform code to divide and encode the payload into several blocks such that the receiver can retain part of the coded information in the corrupted packets. In addition, CH-MAC utilizes hopping to transmit different blocks of a packet with various channels to avoid sudden noise collision. Moreover, CH-MAC adopts a dynamic packet length adjustment mechanism to mitigate network congestion. Extensive evaluations on a real-world hardware testbed and a simulation platform show that CH-MAC can reduce end-to-end transmission latency by 2.63× with a communication success rate requirement of >95% compared with state-of-the-art methods.

As an emerging Low-Power Wide Area Networks (LPWAN) technology, LoRa is dedicated to providing long-range connections for pervasive Internet-of-Things devices. As LoRa operates in the unlicensed spectrum, transmissions from multiple LoRa end-devices inevitably collide into each other, leading to packet losses and increased transmission delay. Targeting at collisions caused by interferences under the same spreading factor (SF) settings, researchers introduce multiple lines of techniques. Despite their efforts, these techniques commonly neglect the potential collisions caused by interferences under different SF settings, which are resulted by the imperfect orthogonality. Given the disparate transmission power configurations and diverse deployed locations, the collisions under different SFs commonly exist in practical networks, and significantly limit the LoRa reliability. In this paper, we present X-MAC, the first scheduler that is aware of imperfect orthogonality. Technically, X-MAC detects the collisions under different SFs via tracking historical transmissions, and further performs dynamic channel scheduling to avoid collisions caused by interferences both under the same and different SFs. Extensive evaluations on testbed devices show that, compared with the state-of-the-art methods, X-MAC boosts the network scalability (number of concurrent end-devices) by 2.41× with packet reception rate (PRR) requirement of > 95%.

Low-Power Wide Area Networks (LPWANs) are widely used to connect large-scale Internet of Things (IoT) applications. Long Range (LoRa) is a promising LPWAN technology sensitive to energy consumption, since LoRa nodes are generally battery-powered, and the battery life will influence the lifetime of the LoRa network. In practice, the battery life of LoRa nodes is short in many scenarios, due to the long transmission distance form the gateway leading to high energy consumption. Existing techniques for energy-efficient data transmission mainly focus on static gateways, and will consume huge energy of remote nodes. In this paper, we propose to integrate LoRa with mobility to minimize the energy consumption of nodes by effectively shortening the transmission distances, and design the first mobile LoRa data transmission system called LoRaDrone by leveraging the unmanned aerial vehicle (UAV) gateway flying close to nodes. Specifically, we present a low-power communication mechanism and a dynamic channel allocation policy to minimize the energy consumed in sensing and communicating with the UAV gateway, while considering the distinctive LoRa parallel reception and complex transmission collisions. Then, an optimal speed scheduling strategy is designed to ensure the reliability of data transmission, and minimize the energy consumption of the UAV. Evaluations on various scales verify the effectiveness of LoRaDrone under different nodes' distributions and UAV paths. Compared with the baselines, the energy consumption of nodes using LoRaDrone is at most reduced by$\mathbf{70.37}\times$at 5000 nodes.

Accurate identification and early diagnosis of malignant lung nodules are crucial for improving the survival rate of patients with lung cancer. Deep learning methods have recently been proven success in computer-aided diagnostic tasks. However, to the best of our knowledge, the features of tissues and vessels will disturb the model resulting in inaccurate classification of the nodules. To reduce the interference and capture crucial contextual information from different channels in a more efficient way, we introduce a Hyper-Attention Mechanism(HAM) that can be easily integrated into convolutional neural networks(CNNs). Moreover, without incorporating prior-domain knowledge, traditional methods lack interpretability, which is difficult to understand and utilize them in the clinic by radiologists. Based on this, we propose a novel Knowledge-Guided model to predict malignant pulmonary nodules from chest CT data, which inject external medical knowledge into CNNs to guide the training process. We evaluate the proposed model on the LIDC-IDRI dataset and demonstrate its effectiveness by achieving comparable state-of-the-art performance.

We present a novel end-to-end deep metric learning model named Class-Prototype Discriminative Network (CPDN) for Generalized Zero-Shot Learning (GZSL). It consists of a generative network for producing the visual prototype of each class by feeding its semantic representation, and a metric network for measuring the similarities between the sample and the generated class-prototypes to accomplish the classification. In CPDN, a query sample intends to posses a higher similarity with its homogenous class-prototypes while the lower similarities with the inhomogenous ones, and the class-prototypes also intend to be distinguished with each other through the metric network. Moreover, a discriminative version of Relation Network (RN) named Discriminative Relation Network (DRN) is presented by incorporating the aforementioned idea into the conventional RN model for further achieving the complementation CDPN and RN in metric learning. Extensive experimental results on standard benchmarks demonstrate that our proposed approaches consistently outperform RN, and achieve the competitive performances compared with the state-of-the-arts in GZSL.