Crowd counting is an important vision task, which faces challenges on continuous scale variation within a given scene and huge density shift both within and across images. These challenges are ...typically addressed using multi-column structures in existing methods. However, such an approach does not provide consistent improvement and transferability due to limited ability in capturing multi-scale features, sensitivity to large density shift, and difficulty in training multi-branch models. To overcome these limitations, a Single-column Scalere-invariant Network (ScSiNet) is presented in this paper, which extracts sophisticated scale-invariant features via the combination of interlayer multi-scale integration and a novel intralayer scale-invariant transformation (SiT). Furthermore, in order to enlarge the diversity of densities, a randomly integrated loss is presented for training our single-branch method. Extensive experiments on public datasets demonstrate that the proposed method outperforms state-of-the-art approaches in counting accuracy and achieves remarkable transferability and scale-invariant property.
Multi-view clustering aims at integrating the complementary information between different views so as to obtain an accurate clustering result. In addition, the traditional clustering is a kind of ...unsupervised learning method, which does not take the label information into learning. In this paper, we propose a novel model, called semi-supervised multi-view clustering based on orthonormality-constrained nonnegative matrix factorization (MVOCNMF), to cluster the multi-view data into a number of categories. In the proposed model, based on the label information, we first learn the low-dimensional representations of data by the constrained NMF technique, and simultaneously cluster the samples with the same label into the clustering prototypes for each view. After that, we put forward a novel orthonormality constraint term to obtain the desirable representations for each view, and use the co-regularization to integrate the complementary information from different views. We further develop an alternating minimization algorithm to solve the proposed model, and present the convergence analysis and computational complexity of the proposed method. Extensive experimental results on several multi-view datasets have shown that the proposed MVOCNMF method outperforms the existing multi-view clustering methods.
Extranodal natural killer (NK)/T cell lymphoma (ENKTL) is an aggressive non-Hodgkin's lymphoma with high mortality and poor prognosis despite radiotherapy and chemotherapy. The current analysis aimed ...to assess the pathological features, clinical features, and prognostic indicators of ENKTL.
120 ENKTL patients were analyzed for pathologic diagnosis and clinical disease manifestations from April 2007 to October 2012. Complete remission, 2-year overall survival, and progression-free survival were analyzed.
Compared with the nasal group, a greater percentage of patients in the non-nasal group intended to receive autologous stem cell transplantation had Epstein-Barr virus (EBV) DNA, Ann Arbor stage IV, Ki-67 expression ≥ 60%, and abnormal ferroprotein and β-microglobulin levels. The rate of complete remission in the non-nasal group was higher than that in the nasal group. The overall survival rate was 74.9% at 24 months. Patients receiving chemotherapy and radiotherapy were more likely to have disease progression compared with patients who received chemotherapy or radiotherapy alone.
Further understanding the pathological and clinical features of ENKTL will be critical for moving forward. Ki-67, β-microglobulin, EBV DNA, and primary site prognostic indicators may be useful to stratify patients into different risk groups, to gain insight into patient-specific treatments, and to potentially improve survival.
AbstractFe3O4-coated pretreated biochars (Fe3O4@PBC) were prepared for the first time by a Fe(III)-ethanol solution impregnation-calcination method. When photo-Fenton catalysts were used, their ...effectiveness in removing metronidazole (MNZ) from aqueous media was evaluated. Fe3O4@PBC samples were characterized by X-ray diffraction, scanning electron microscope, vibrating sample magnetometer, X-ray photoelectron spectroscopy, and Brunauer–Emmett–Teller methods. The results showed that Fe3O4 coating was successfully formed on the surface of HNO3- pretreated biochar, and Fe3O4@PBC can be separated by applying an external magnetic field. The coating of Fe3O4 did not change the pore structure and maintained a high surface area of the biochar. Fe loading significantly affected the photo-Fenton degradation and adsorption ability of MNZ. The highest MNZ removal rate and the greatest catalytic ability were found in the PBC-6.6Fe sample containing 6.6% Fe by mass. Various operating parameters, such as solution pH, H2O2 concentration, and MNZ concentration, were tested during MNZ’s photo-Fenton catalytic degradation. The results indicate that the highest MNZ degradation efficiency can be derived from a moderate acidic solution, and the optimal pH is 3. Using PBC-6.6Fe, the increase of H2O2 concentration from 30 to 60 mmol·L−1 promotes the degradation of photo-Fenton, and both an excessive H2O2 and an increase in MNZ concentration suppressed the process. Under the conditions of 0.4 g·L−1 PBC-6.6Fe, 300 mg·L−1 MNZ, 60 mmol·L−1 H2O2, and initial pH of 3, 95.1% of MNZ was degraded. The PBC-6.6Fe had good stability, and its removal efficiency was still over 92% after five repeated uses. This study confirmed that •OH played a dominant role, while O2•− and h+ played a weaker role in the photo-Fenton system. The results indicated that Fe3O4@PBC served as a prospective visible-light-driven catalyst similar to Fenton for the treatment of wastewater containing MNZ.
Improving power conversion efficiency of photovoltaic devices has been widely investigated; however, most research studies mainly focus on the modification of the absorber layer. Here, we present an ...approach to enhance the efficiency of Cu(In,Ga)(S,Se)2 (CIGSSe) thin-film solar cells simply by tuning the CdS buffer layer. The CdS buffer layer was deposited by chemical bath deposition. Indium doping was done during the growth process by adding InCl3 into the growing aqueous solution. We show that the solar cell efficiency is increased by proper indium doping. Based on the characteristics of the single CdS (with or without In-doping) layer and of the CIGSSe/CdS interface, we conclude that the efficiency enhancement is attributed to the interface-defect passivation of heterojunction, which significantly improves both open circuit voltage and fill factor. The results were supported by SCAPS simulations, which suggest that our approach can also be applied to other buffer systems.
Tripolinolate A (TLA) is recently identified as a new compound from a halophyte plant Tripolium vulgare and has been shown to have significant in vitro activity against the proliferation of ...colorectal cancer and glioma cells. This study was designed to further investigate the effects of TLA on the proliferation of human normal cells, and the apoptosis and cell cycle in colorectal cancer cells, and the growth of tumors in the colorectal cancer-bearing animals. The data obtained from this study demonstrated that: 1) TLA had much less cytotoxicity in the human normal cells than the colorectal cancer cells; 2) TLA remarkably induced apoptosis in the human colorectal cancer cells and blocked cell cycle at G2/M phase, and 3) TLA had significant anti-colorectal cancer activity in the tumor-bearing animals.
A hybrid design proposal for nonvolatile (NV) SRAM bit-cell is implemented in this work. Magnetic tunnel junction (MTJ) and Tunnel field effect transistor (TFET) are hierarchically integrated for ...ultra low leakage (ULL) and low supply voltage (V dd ) operations. Typical NV-SRAM structures are investigated with TFET replacement. Utilizing AlGaSb/InAs broken-gap heterojunctions, a novel NV-TFET-SRAM with 8T1M structure is proposed. A sub-0.5V SRAM operations and 1.65pW leakage power are realized in this hybrid bit-cell, fulfilling the needs of ultra-low dynamic power and ultra-low-leakage Internet-of-things (IoT) applications. Operation modes including normal, store, restore and reset are configured in the MTJ/TFET based 8T1M bit-cell, for writing/sensing operations in both volatile (SRAM) and non-volatile memory (MTJ). MOSFET base sensing amplifiers are evaluated with TFET implementation, for further research of more energy efficient and reliable MTJ reading.
We have developed a simple and cost-effective one-step surface oxidation method to synthesize three-dimensional (3D) flower-like, similar to that of a blooming chrysanthemum, CuO hierarchical ...nanostructures directly on a copper foam, which is acting as the Cu source and the current collector. The as-prepared sample can be directly used as a binder-free electrode for supercapacitors. Benefiting from the novel synthesis strategy and the 3D connect/quasi-connect structures, the as-prepared CuO/copper foam electrode can provide massive active sites for redox reactions, high electronic conductivity, short diffusion pathway for ions and effectively electrolyte penetrating. These characteristics together with the synergy effect between CuO and copper foam substrate lead to a high capacitance of 1641.4mFcm−2, good rate capability (77.2% retention upon increasing the current density by 10 times) and good cyclability (79% retention after 10000 cycles). These results suggest that the CuO electrode demonstrates a good potential for high performance energy storage devices.
We propose a sub-<inline-formula> <tex-math notation="LaTeX">\mu \text{W} </tex-math></inline-formula> always-ON keyword spotting (<inline-formula> <tex-math notation="LaTeX">\mu ...</tex-math></inline-formula>KWS) chip for audio wake-up systems. It is mainly composed of a neural network (NN) and a feature extraction (FE) circuit. For significantly reducing the memory footprint and computational load, four techniques are used to achieve ultra-low-power consumption: 1) a serial-FFT-based Mel-frequency cepstrum coefficient circuit is designed for FE, instead of the common parallel FFT. 2) A small-sized binarized depthwise separable convolutional NN (DSCNN) is designed as the classifier. 3) A framewise incremental computation technique is devised in contrast to the conventional whole-word processing. 4) Reduced computation allows a low system clock frequency, which enables near-threshold voltage operation, and low leakage memory blocks are designed to minimize the leakage power. Implemented in 28-nm CMOS technology, this <inline-formula> <tex-math notation="LaTeX">\mu </tex-math></inline-formula>KWS consumes <inline-formula> <tex-math notation="LaTeX">0.51~\mu \text{W} </tex-math></inline-formula> at a 40-kHz frequency and a 0.41-V supply, with an area of 0.23 mm 2 . Using the Google speech command data set, 97.3% accuracy is reached for a one-word KWS task and 94.6% for a two-word task.
Speech disorders are associated with different degrees of functional and structural abnormalities. However, the abnormalities associated with specific disorders, and the common abnormalities shown by ...all disorders, remain unclear. Herein, a meta-analysis was conducted to integrate the results of 70 studies that compared 1843 speech disorder patients (dysarthria, dysphonia, stuttering, and aphasia) to 1950 healthy controls in terms of brain activity, functional connectivity, gray matter, and white matter fractional anisotropy. The analysis revealed that compared to controls, the dysarthria group showed higher activity in the left superior temporal gyrus and lower activity in the left postcentral gyrus. The dysphonia group had higher activity in the right precentral and postcentral gyrus. The stuttering group had higher activity in the right inferior frontal gyrus and lower activity in the left inferior frontal gyrus. The aphasia group showed lower activity in the bilateral anterior cingulate gyrus and left superior frontal gyrus. Across the four disorders, there were concurrent lower activity, gray matter, and fractional anisotropy in motor and auditory cortices, and stronger connectivity between the default mode network and frontoparietal network. These findings enhance our understanding of the neural basis of speech disorders, potentially aiding clinical diagnosis and intervention.