In this paper, we propose a novel single image super-resolution (SR) method based on low-rank sparse representation with self-similarity learning. Sparse representation is known as a promising method ...for SR. However, the sparse codes for low resolution (LR) patches gained by conventional method are not faithful to those for the original high resolution (HR) ones. To overcome this defect, we explore the structures of sparse representation for nonlocal similar patches in natural images by low-rank strategy. It assumes that the sparse codes for nonlocal similar patches should be low-rank. By low-rank constraint, similar components of sparse codes are shared and coding noises are removed, which improves coding accuracy and SR performance. Furthermore, we utilize self-similarity learning framework to generate a self-examples dictionary compatible to the low-rank sparse representation based SR. Experimental results demonstrate that our proposed method can recover good SR results both quantitatively and perceptually.
In this work, a memristive spike-based computing in memory (CIM) system with adaptive neuron (MSPAN) is proposed to realize energy-efficient remote arrhythmia detection with high accuracy in edge ...devices by software and hardware co-design. A multi-layer deep integrative spiking neural network (DiSNN) is first designed with an accuracy of 93.6% in 4-class ECG classification tasks. Then a memristor-based CIM architecture and the corresponding mapping method are proposed to deploy the DiSNN. By evaluation, the overall system achieves an accuracy of over 92.25% on the MIT-BIH dataset while the area is 3.438 mm
and the power consumption is 0.178 μJ per heartbeat at a clock frequency of 500 MHz. These results reveal that the proposed MSPAN system is promising for arrhythmia detection in edge devices.
Budd-Chiari Syndrome (BCS) is a relatively rare clinical disorder with a wide range of symptoms, caused by the obstruction of the hepatic venous outflow. The etiology and pathogenesis of BCS vary in ...different countries and regions. In Western countries, hepatic venous obstruction is the most common type, and its main cause is closely related to the hypercoagulable state of the body. Inferior vena cava obstruction is common in Asia, and its etiology progresses slowly due to the lack of epidemiological data. 3 Here, we report a rare case of BCS associated with the hypereosinophilic syndrome and discuss the possible causal relationship between the two.
The patient was a 33-year-old female with intermittent epistaxis, gum bleeding, and excessive menstrual flow for the past 6 months. The routine blood tests showed elevated levels of eosinophils, and the liver function test showed mildly elevated levels of γ-glutamyl transpeptidase and alkaline phosphatase, and abdominal ultrasound showed hepatosplenomegaly and suspicion of intrahepatic arteriovenous or arteriovenous-portal fistula.
Finally, through the improvement of bone marrow aspiration, digital subtraction angiography and gene detection, the diagnosis of BCS combined with hypereosinophilic syndrome was confirmed, and JAK2V617F mutation was highly associated with it.
The patient received endovascular stent implantation and regular oral rivaroxaban anticoagulation therapy after operation.
Seven months later, enhanced computed tomography (CT) of the hepatobiliary showed that the hepatic bruise-like changes were significantly reduced compared with before, and the right hepatic vein and the right perihepatic vein stent were left in place with a good filling of contrast in the stent.
The patient, in this case, was finally diagnosed with BCS combined with hypereosinophilic syndrome, and to our knowledge, such case reports are rare. Our case report suggest an association between BCS and hypereosinophilic syndrome, but relevant studies are minimal, we hope to conduct larger and higher quality studies on these patients in the future, to provide new directions and basis for the etiology and pathogenesis of these diseases, as well as provide new targets and ideas for clinical treatment.
The ubiquity of edge devices has led to a growing amount of unlabeled data produced at the edge. Deep learning models deployed on edge devices are required to learn from these unlabeled data to ...continuously improve accuracy. Self-supervised representation learning has achieved promising performances using centralized unlabeled data. However, the increasing awareness of privacy protection limits centralizing the distributed unlabeled image data on edge devices. While federated learning has been widely adopted to enable distributed machine learning with privacy preservation, without a data selection method to efficiently select streaming data, the traditional federated learning framework fails to handle these huge amounts of decentralized unlabeled data with limited storage resources on edge. To address these challenges, we propose a Self-supervised On-device Federated learning framework with coreset selection, which we call SOFed, to automatically select a coreset that consists of the most representative samples into the replay buffer on each device. It preserves data privacy as each client does not share raw data while learning good visual representations. Experiments demonstrate the effectiveness and significance of the proposed method in visual representation learning.
This article presents a wideband, low-jitter frequency synthesizer utilizing a dual-mode voltage-controlled oscillator (VCO). Mode imbalance in the dual-mode VCO is analyzed theoretically and ...compensated through the proposed symmetric figure-8 transformer and capacitor arrays. The compact mode-switching circuitry fundamentally eliminates mode ambiguity in multi-mode autonomous circuits. A computer-aided algorithm based on sequential least-squares programming (SLSQP) and hierarchical optimization method is developed to automatically optimize the capacitor array in the wideband VCO. The implemented dual-mode VCO suppresses the phase noise (PN) difference across the operating frequency range, which further enables a sub-sampling phase-locked loop (SSPLL) to achieve near-minimum jitter across a wide frequency range without loop gain adaptation. Fabricated in a 40-nm CMOS process, the wideband SSPLL covers the frequency range of 7.9-14.3 GHz with 14.1-17.2-mW power consumption and occupies only 0.18-mm 2 area. The SSPLL achieves better than −115-dBc/Hz in-band PN at a 10-GHz carrier. The rms jitter is less than 85 fs across the whole frequency range. The corresponding figure-of-merit tuning (FoM<inline-formula> <tex-math notation="LaTeX">_{T} </tex-math></inline-formula>) is −247.1 to −248.1 dB.
In this paper, a circularly polarized rectenna is proposed for microwave energy transmission, which is composed of a broadband Substrate Integrated Waveguide (SIW) antenna and a rectifier circuit. A ...dielectric resonator is loaded on the SIW antenna, enhancing the impedance bandwidth. Circular polarization is achieved by chamfering the diagonal of the rectangular patch. Two sections of microstrip branches are employed for good impedance matching over the broadband operation. Measured results show that the proposed rectenna has a good impedance performance (|S11| < −10 dB) and high RF-DC conversion efficiency (>50 %) over a relative bandwidth of 22.9 % (29–36.5 GHz). The maximum conversion efficiency of 67.5 % is realized at the input power of 18 dBm.
Full text
Available for:
IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Human-oriented image captioning with both high diversity and accuracy is a challenging task in vision+language modeling. The reinforcement learning (RL) based frameworks promote the accuracy of image ...captioning, yet seriously hurt the diversity. In contrast, other methods based on variational auto-encoder (VAE) or generative adversarial network (GAN) can produce diverse yet less accurate captions. In this work, we devote our attention to promote the diversity of RL-based image captioning. To be specific, we devise a partial off-policy learning scheme to balance accuracy and diversity. First, we keep the model exposed to varied candidate captions by sampling from the initial state before RL launched. Second, a novel criterion named max-CIDEr is proposed to serve as the reward for promoting diversity. We combine the above-mentioned offpolicy strategy with the on-policy one to moderate the exploration effect, further balancing the diversity and accuracy for human-like image captioning. Experiments show that our method locates the closest to human performance in the diversity-accuracy space, and achieves the highest Pearson correlation as 0.337 with human performance.