This paper presents a fully integrated NMOS stimulator using a hybrid dynamic bulk biasing technique (HDBT) and a charge-pump-like control technique (CCT) in a 180-nm bulk CMOS technology. HDBT integrates terminal-voltage-dependent and logic dynamic bulk biasing to set the bulk bias voltage according to the electrode voltage. CCT adds a DC voltage to the gate terminal through a diode and capacitor to help turn on the NMOS transistor. It helps turn off the transistor by shorting the source and gate terminals together and applying two diodes across the drain and source terminals. To achieve an electrode voltage higher than the breakdown voltage of substrate diode ($V_{\mathrm {BD}}$) with an independent power supply, a high voltage tolerant switch is proposed with HDBT and CCT. A high voltage interface is also proposed, utilizing the capacitor adaptive biasing, to overcome the limitation of$V_{\mathrm {BD}}$between the high and low voltage domains and to accommodate the variation of electrode voltage. Fabricated in a 180-nm standard CMOS technology, the stimulator achieves a maximum electrode voltage ($V_{\mathrm {E,MAX}}$) of 18.74V under a 3.3-V supply, with a highest$V_{\mathrm {E,MAX}}$/$V_{\mathrm {BD}}$ratio of 1.27 than state-of-the-art stimulators, including non-standard technology designs. In a continuous output test mode over 10million cycles, the variation of$V_{\mathrm {E,MAX}}$is less than 150mV. The measured maximum residual voltage on the capacitor is 13.55mV.

This brief presents an area-efficient noise-shaping (NS) successive approximation register (SAR) analog-to-digital converter (ADC) employing a parallel-delayed sampling (PDS) technique. PDS samples the residual voltages from multiple ADC conversion cycles to increase the NS effect, without the need for large integration capacitors of the typical cascaded passive integrators. A preamplifier is placed between the sampling capacitors and the integrator to avoid signal attenuation, while further reducing the area of the integrator. PDS and preamplifier introduce two left-half-plane poles to the noise transfer function (NTF) to boost the NS effect, while reducing the impact of the parasitic capacitance to essentially enhance the robustness. A prototype 9-bit NS-SAR ADC is designed in a 130-nm CMOS process. At an oversampling ratio (OSR) of 16, the proposed PDS NS-SAR ADC achieves 80.93-dB peak signal to noise and distortion ratio (SNDR) and provides 4.2 NS/area efficiency factor. It consumes a power of$23.46~\mu $W over a bandwidth of 19.53 kHz, achieving a Schreier figure of merit (FoM${}_{\mathrm {S}}$) of 170.13 dB.

This brief presents a CMOS inverter (INV)-based amplifier with local and system dual common-mode feedback (CMFB) loops through paralleled pseudo transconductors (PTs) for multichannel neural signal recording. The PT-INV forms a DC-coupled input to ensure a high input impedance, and local and system dual CMFB loops are introduced through the paralleled PT terminals of the INV. The local CMFB (L-CMFB) with the capacitor-reused topology not only reduces the die area and increases the differential-mode (DM) gain but also reduces the common-mode (CM) gain. The system CMFB (S-CMFB) loop detects the multiple CM outputs, and further increases the overall CMRR. The proposed PT-INV based instrumentation amplifiers (IAs) with dual CMFB loops are fabricated in a 0.18-$\mu $m CMOS process, and an overall CMRR of 87dB is achieved with a compact single-channel core area of 0.02mm2.

ABSTRACT Object detection in remote sensing imagery requires precise localisation and identification of targets under challenging conditions. Facing the challenges of arbitrary target orientations, wide‐scale variations, dense distributions, and small objects in remote sensing object detection, anchor‐based methods suffer from inadequate rotated target representation using rectangular boxes. This necessitates excessive angle‐specific anchors, leading to heavy computational overhead, severe sample imbalance, and slow speeds unsuitable for mobile deployment. To address these accuracy‐efficiency trade‐offs, we propose DSSNet: an anchor‐free rotated object detection network with dynamic sample selection for remote sensing images. DSSNet replaces traditional backbones with the parameter‐efficient ConvNeXt‐T and utilises an FPN for accelerated multi‐scale feature extraction. During prediction, it employs a shape‐adaptive selection strategy combined with a contour point quality assessment strategy to dynamically refine target contour points, enabling real‐time rotated object detection. The efficacy of DSSNet has been thoroughly validated through benchmark comparisons on diverse datasets. On the DOTA dataset, DSSNet clearly outperforms baseline methods in detection performance, achieving a mean Average Precision (mAP) of 76.97% and the fastest detection speed of 26.2 frames per second (FPS).

An oscillator-based reconfigurable modulator is proposed to support multi-mode and fast modulation. A direct-modulation structure composed of the cross-coupled oscillator with the fast-switched film bulk acoustic resonator (FBAR) is used to enhance the frequency stability under fast OOK/BFSK modulation. To avoid extra phase-reversal circuitry, a polarity-swapped switching structure is employed in the differential branches of the modulator to achieve energy-efficient BPSK modulation, and this structure is also reused as a buffer stage for OOK/BFSK modulation to avoid the loading effect. In addition, an adaptive fast-switching technique is also proposed to improve OOK/BFSK modulation data rate and energy efficiency. The modulator is fabricated in a 180 nm CMOS technology. The free-running oscillation frequencies with two FBARs are 962 MHz and 990 MHz, and the measured phase noises are -137.3 dBc/Hz@1MHz and -137.1 dBc/Hz@1MHz, respectively. For OOK/BFSK/BPSK modulation, the proposed modulator demonstrated 280/325/67.6 pJ/bit energy efficiency and 5.63/4.20/5.55 % rms EVM with 10/10/50 Mbps data rates.

This article presents a Gm-controlled inverter (GC-INV) based amplifier with multiple common-mode feedback (CMFB) loops for multi-channel bio-signal recording. The GC-INV forms a DC-coupled input to ensure a high input impedance. The multi-CMFB, including twin local (TL), regional system (RS), and averaged system (AS) CMFB loops, is introduced through the paralleled GC terminals to provide multiple feedback paths. The TL-CMFB with capacitor-reused topology not only reduces the die area and increases the differential-mode gain, but also reduces the common-mode (CM) gain. The RS-CMFB mitigates the common-mode interference (CMI) due to the mismatch of the CM feedback paths. The AS-CMFB further mitigates the accumulated CMI from CM sampling paths. These CMFB loops avoid the design trade-off between the intrinsic CMRR and the efficiency of area and power. The proposed GC-INV based amplifier with multi-CMFB is fabricated in a 0.18-$\mu $m CMOS technology. It achieves an intrinsic CMRR of 97 dB, TCMRR of 78 dB, and the single-channel INV consumes a chip area of 0.008 mm2.

Recently, GNN(Graph Neural Network) recommender systems have benefitted from contrastive learning as an auxiliary task of recommendation and have employed data augmentation to overcome the data sparsity problem. However, we find that the training process of contrastive learning is affected by the popularity bias due to the longtail distribution of interaction data, resulting in the inadequate feature training of low-degree nodes. To address this problem, we propose DCLGCF (Debiased Contrastive Learning For Graph Collaborative Filtering). More specifically, we propose two data augmentation methods with respect to popularity reduction and longtail enhancement. In addition, we propose Mixed-InfoNCE, which designs a novel mixed sampling strategy and introduce a new contrastive learning loss function by considering a frequency penalty term, aiming at increasing the contribution of longtail items to the gradient calculation, and enhancing the training of longtail item features. To validate the effectiveness of our proposed DCLGCF, we conduct thorough experiments on four real-world datasets. The results clearly demonstrate that DCLGCF outperforms existing models in terms of recommendation accuracy, and remarkable improvements are achieved especially when recommending longtail items.

In this paper, a voltage-controlled oscillator (VCO) and phase detector (PD) based stimulation artifact (SA) removal scheme is proposed for simultaneous neural stimulation and recording. VCO is used to avoid the saturation of conventional front-end amplifier. The VCO converted signals are separated in the frequency and time domain. A PD in closed- loop recovers the neural signal from the converted signal, and a predicted SA amplitude helps removing the SA. Compared to other SA removal schemes, the proposed VCO-PD-based SA removal technique avoids the subtraction path at the input stage to ensure a high input impedance and achieves high linearity for continuous SA removal.

BACKGROUND: This study aimed to investigate the characteristics of gray matter volume loss in hippocampal subfields and alterations in whole-brain functional connectivity among patients with diabetic retinopathy (DR), and to examine the correlations between these neural changes and neuropsychological scale scores as well as clinical indicators. METHODS: Structural and functional magnetic resonance imaging (MRI) data, along with neurocognitive assessments, were acquired from 32 patients with diabetic retinopathy (DR) and 38 age- and sex-matched healthy controls using a Siemens 3.0T MRI scanner. Hippocampal subfield volumes were segmented and extracted using the Anatomy toolbox and Restplus (v1.25) within the MATLAB R2022a environment. Subregions demonstrating significant volumetric alterations were identified through subsequent between-group comparisons and defined as regions of interest (ROIs) for seed-based whole-brain functional connectivity (FC) analysis. The relationships between gray matter volume, functional connectivity values, neuropsychological test scores, and clinical indices were investigated using partial correlation analysis. RESULTS: Compared to healthy controls, patients with diabetic retinopathy (DR) exhibited significantly reduced bilateral entorhinal cortex (EC) volumes and increased volume in bilateral dentate gyrus (DG) (P < 0.05, FDR-corrected). Using these volume-altered subregions as regions of interest (ROIs), seed-based functional connectivity (FC) analysis revealed significantly increased FC strength between the left EC and the left caudate nucleus, alongside significantly decreased FC strength between the right DG and the right middle occipital gyrus (GRF correction: voxel-level p < 0.001, cluster-level p < 0.05, two-tailed). Furthermore, partial correlation analyses demonstrated significant positive correlations between the FC strength of the right hippocampal DG subfield and the right middle occipital gyrus, and scores on both the Montreal Cognitive Assessment (MoCA) (r = 0.560, p = 0.001) and the Mini-Mental State Examination (MMSE) (r = 0.541, p = 0.002). CONCLUSIONS: Patients with diabetic retinopathy (DR) demonstrate structural alterations in hippocampal subfields and widespread functional dysconnectivity across the brain. Notably, the strength of functional connectivity between the hippocampus and the visual cortex was significantly and positively correlated with cognitive function. This study provides neuroimaging evidence supporting the mechanisms underlying central nervous system complications in DR, suggesting that hippocampal subfield volumes and specific functional connectivity patterns hold potential as neuroimaging biomarkers for early intervention.

The development of creative applications and intelligent gadgets requires a secure and straightforward interface with human users. We propose, design, and implement ThumbUp, a smartwatch-based two-factor real-time identification and authentication system in which smartwatch users can identify and authenticate themselves using some simple hand and finger movements, such as thumb-up. ThumbUp leverages the signal from the Inertial Measurement Unit (IMU) in in Commercial-Off-The-Shelf (COTS) smart devices to discover the unique pattern generated by each user's simple gestures using a carefully constructed deep learning model. Smart homes provide a comfortable, safe, and efficient living environment, epecially help the sick and aged. We propose strategies for convenient and reliable control in smart homes with gesture command recognition. We build an Auto-Encoder-based filter that reconstructs the raw data to improve the representation of gesture features. Moreover, we adopt the random forest method to analyse the contextual command correlation. And we employ the authentication system based on smartwatch for personalized command feedback and ensure that illegals cannot use the device. We implement our system and undertake rigorous studies to determine its usefulness and efficiency over a three-month period with 65 users. It achieves a 97% accuracy for user classification and an EER of 0.014 for authentication task with a single simple gesture. And our method achieves 91% accuracy for command recognition and 96% command accuracy with contextual informations. Additionally, we conduct a study of user acceptability of our system and explain how gesture proficiency influences authentication accuracy.

This brief proposes a parallel multiresidual (PMR) integrator to enhance the noise-shaping (NS) effect for successive approximation register (SAR) analog-to-digital converter (ADC). The PMR employs passive integrators in parallel to simultaneously integrate the average result of the multiple sequential residual voltages. The proposed PMR technique provides an alternative scheme to enhance the NS rather than increasing the order of the integrator to suppress the instability and power. A prototype 7-bit second-order NS-SAR ADC is designed and simulated in a 130-nm CMOS process. PMR increases the effective number of bits (ENOBs) to 10.6 bit, which enhances the NS effect of 3.6 bit. It achieves a peak signal-to-noise and distortion ratio (SNDR) of 65.84 dB over a bandwidth of 1.3 kHz at the oversampling ratio (OSR) of 16.

The widespread creative application and smart devices call for convenient and secure interaction with human users. We propose, design, and implement a smartwatch-based two-factor real-time identification and authentication system named ThumbUp, where smartwatch users can identify and authenticate themselves by some simple hand and finger gestures, such as thumb-up. ThumbUp leverages the signal collected from the Inertial Measurement Unit (IMU) in Commercial Off-The-Shelf (COTS) smart devices and discovers the unique fingerprint pattern produced by each user’s simple hand gestures using a carefully crafted deep learning model. We implement our system and conduct extensive experiments to evaluate its efficacy and efficiency with 65 different users over a period of more than 3 months. It reaches an accuracy of 97% for identification, and EER 0.014 for authentication using only one simple gesture. We also survey the users’ acceptance of our system and discuss how the proficiency of gestures affects authentication accuracy.

To further improve the convenience and effectiveness of human computer interaction (HCI) with smart devices, human activity recognition (HAR) has been widely studied from various aspects. Unfortunately, deep learning based methods often suffer from either expensive labeling efforts or weak generalization ability. Inspired by recently developed domain adaptation strategies, we propose XHAR, a novel adversarial deep domain adaptation framework for HAR using smart devices, providing better device and user adaptation. XHAR first selects the most similar source dataset (with label), then extracts device and user independent spatial-temporal features through the combinations of Convolutional Neural Networks (CNN) and Bidirectional Gated Recurrent Units (BiGRU) feature extractors. Moreover, it removes the distribution discrepancy using multiple domain discriminators, and finally performs adaptation on the target dataset (without label) to obtain the predicted labels. We conduct extensive experiments on 50 users (i.e., of different ages, genders, and body shapes) and 4 smart devices with two kinds of datasets (i.e., gesture activities and sport activities). We compare our method with the source-only model and several state-of-the-art domain adaptation models. The results show that XHAR increases the classification accuracy by at least 4.81% (to 74.50%) on the adaptation between different users, and accordingly by at least 9.25% (to 69.23%) between different devices.