We describe a method for articulated human detection and human pose estimation in static images based on a new representation of deformable part models. Rather than modeling articulation using a ...family of warped (rotated and foreshortened) templates, we use a mixture of small, nonoriented parts. We describe a general, flexible mixture model that jointly captures spatial relations between part locations and co-occurrence relations between part mixtures, augmenting standard pictorial structure models that encode just spatial relations. Our models have several notable properties: 1) They efficiently model articulation by sharing computation across similar warps, 2) they efficiently model an exponentially large set of global mixtures through composition of local mixtures, and 3) they capture the dependency of global geometry on local appearance (parts look different at different locations). When relations are tree structured, our models can be efficiently optimized with dynamic programming. We learn all parameters, including local appearances, spatial relations, and co-occurrence relations (which encode local rigidity) with a structured SVM solver. Because our model is efficient enough to be used as a detector that searches over scales and image locations, we introduce novel criteria for evaluating pose estimation and human detection, both separately and jointly. We show that currently used evaluation criteria may conflate these two issues. Most previous approaches model limbs with rigid and articulated templates that are trained independently of each other, while we present an extensive diagnostic evaluation that suggests that flexible structure and joint training are crucial for strong performance. We present experimental results on standard benchmarks that suggest our approach is the state-of-the-art system for pose estimation, improving past work on the challenging Parse and Buffy datasets while being orders of magnitude faster.
This paper focuses on the unsupervised domain adaptation of transferring the knowledge from the source domain to the target domain in the context of semantic segmentation. Existing approaches usually ...regard the pseudo label as the ground truth to fully exploit the unlabeled target-domain data. Yet the pseudo labels of the target-domain data are usually predicted by the model trained on the source domain. Thus, the generated labels inevitably contain the incorrect prediction due to the discrepancy between the training domain and the test domain, which could be transferred to the final adapted model and largely compromises the training process. To overcome the problem, this paper proposes to explicitly estimate the prediction uncertainty during training to rectify the pseudo label learning for unsupervised semantic segmentation adaptation. Given the input image, the model outputs the semantic segmentation prediction as well as the uncertainty of the prediction. Specifically, we model the uncertainty via the prediction variance and involve the uncertainty into the optimization objective. To verify the effectiveness of the proposed method, we evaluate the proposed method on two prevalent synthetic-to-real semantic segmentation benchmarks, i.e., GTA5
→
Cityscapes and SYNTHIA
→
Cityscapes, as well as one cross-city benchmark, i.e., Cityscapes
→
Oxford RobotCar. We demonstrate through extensive experiments that the proposed approach (1) dynamically sets different confidence thresholds according to the prediction variance, (2) rectifies the learning from noisy pseudo labels, and (3) achieves significant improvements over the conventional pseudo label learning and yields competitive performance on all three benchmarks.
The sudden outbreak of 2019 novel coronavirus (2019-nCoV, later named SARS-CoV-2) in Wuhan, China, which rapidly grew into a global pandemic, marked the third introduction of a virulent coronavirus ...into the human society, affecting not only the healthcare system, but also the global economy. Although our understanding of coronaviruses has undergone a huge leap after two precedents, the effective approaches to treatment and epidemiological control are still lacking. In this article, we present a succinct overview of the epidemiology, clinical features, and molecular characteristics of SARS-CoV-2. We summarize the current epidemiological and clinical data from the initial Wuhan studies, and emphasize several features of SARS-CoV-2, which differentiate it from SARS-CoV and Middle East respiratory syndrome coronavirus (MERS-CoV), such as high variability of disease presentation. We systematize the current clinical trials that have been rapidly initiated after the outbreak of COVID-19 pandemic. Whereas the trials on SARS-CoV-2 genome-based specific vaccines and therapeutic antibodies are currently being tested, this solution is more long-term, as they require thorough testing of their safety. On the other hand, the repurposing of the existing therapeutic agents previously designed for other virus infections and pathologies happens to be the only practical approach as a rapid response measure to the emergent pandemic, as most of these agents have already been tested for their safety. These agents can be divided into two broad categories, those that can directly target the virus replication cycle, and those based on immunotherapy approaches either aimed to boost innate antiviral immune responses or alleviate damage induced by dysregulated inflammatory responses. The initial clinical studies revealed the promising therapeutic potential of several of such drugs, including
, a broad-spectrum antiviral drug that interferes with the viral replication, and
, the repurposed antimalarial drug that interferes with the virus endosomal entry pathway. We speculate that the current pandemic emergency will be a trigger for more systematic drug repurposing design approaches based on big data analysis.
Abstract
Tuning metal–support interaction has been considered as an effective approach to modulate the electronic structure and catalytic activity of supported metal catalysts. At the atomic level, ...the understanding of the structure–activity relationship still remains obscure in heterogeneous catalysis, such as the conversion of water (alkaline) or hydronium ions (acid) to hydrogen (hydrogen evolution reaction, HER). Here, we reveal that the fine control over the oxidation states of single-atom Pt catalysts through electronic metal–support interaction significantly modulates the catalytic activities in either acidic or alkaline HER. Combined with detailed spectroscopic and electrochemical characterizations, the structure–activity relationship is established by correlating the acidic/alkaline HER activity with the average oxidation state of single-atom Pt and the Pt–H/Pt–OH interaction. This study sheds light on the atomic-level mechanistic understanding of acidic and alkaline HER, and further provides guidelines for the rational design of high-performance single-atom catalysts.
Nowadays, the prevailing detectors of steganographic communication in digital images mainly consist of three steps, i.e., residual computation, feature extraction, and binary classification. In this ...paper, we present an alternative approach to steganalysis of digital images based on convolutional neural network (CNN), which is shown to be able to well replicate and optimize these key steps in a unified framework and learn hierarchical representations directly from raw images. The proposed CNN has a quite different structure from the ones used in conventional computer vision tasks. Rather than a random strategy, the weights in the first layer of the proposed CNN are initialized with the basic high-pass filter set used in the calculation of residual maps in a spatial rich model (SRM), which acts as a regularizer to suppress the image content effectively. To better capture the structure of embedding signals, which usually have extremely low SNR (stego signal to image content), a new activation function called a truncated linear unit is adopted in our CNN model. Finally, we further boost the performance of the proposed CNN-based steganalyzer by incorporating the knowledge of selection channel. Three state-of-the-art steganographic algorithms in spatial domain, e.g., WOW, S-UNIWARD, and HILL, are used to evaluate the effectiveness of our model. Compared to SRM and its selection-channel-aware variant maxSRMd2, our model achieves superior performance across all tested algorithms for a wide variety of payloads.
We report first-principles and strongly correlated calculations of the newly discovered heavy fermion superconductor UTe2. Our analyses reveal three key aspects of its magnetic, electronic, and ...superconducting properties that include (i) a two-leg ladder-type structure with strong magnetic frustrations, which might explain the absence of long-range orders and the observed magnetic and transport anisotropy, (ii) quasi-two-dimensional Fermi surfaces composed of two separate electron and hole cylinders with similar nesting properties as in UGe2, which may potentially promote magnetic fluctuations and help to enhance the spin-triplet pairing, and (iii) a unitary spin-triplet pairing state of strong spin-orbit coupling at zero field, with point nodes presumably on the heavier hole Fermi surface along the kx direction, in contrast to the previous belief of nonunitary pairing. Our proposed scenario is in excellent agreement with latest thermal conductivity measurement and provides a basis for understanding the peculiar magnetic and superconducting properties of UTe2.
In this paper, we propose a novel semisupervised feature selection framework by mining correlations among multiple tasks and apply it to different multimedia applications. Instead of independently ...computing the importance of features for each task, our algorithm leverages shared knowledge from multiple related tasks, thus improving the performance of feature selection. Note that the proposed algorithm is built upon an assumption that different tasks share some common structures. The proposed algorithm selects features in a batch mode, by which the correlations between various features are taken into consideration. Besides, considering the fact that labeling a large amount of training data in real world is both time-consuming and tedious, we adopt manifold learning, which exploits both labeled and unlabeled training data for a feature space analysis. Since the objective function is nonsmooth and difficult to solve, we propose an iteractive algorithm with fast convergence. Extensive experiments on different applications demonstrate that our algorithm outperforms the other state-of-the-art feature selection algorithms.
The exploration of novel systems for the electrochemical CO2 reduction reaction (CO2RR) for the production of hydrocarbons like CH4 remains a giant challenge. Well‐designed electrocatalysts with ...advantages like proton generation/transferring and intermediate‐fixating for efficient CO2RR are much preferred yet largely unexplored. In this work, a kind of Cu‐porphyrin‐based large‐scale (≈1.5 μm) and ultrathin nanosheet (≈5 nm) has been successfully applied in electrochemical CO2RR. It exhibits a superior FECH4
of 70 % with a high current density (−183.0 mA cm−2) at −1.6 V under rarely reported neutral conditions and maintains FECH4
>51 % over a wide potential range (−1.5 to −1.7 V) in a flow cell. The high performance can be attributed to the construction of numerous hydrogen‐bonding networks through the integration of diaminotriazine with Cu‐porphyrin, which is beneficial for proton migration and intermediate stabilization, as supported by DFT calculations. This work paves a new way in exploring hydrogen‐bonding‐based materials as efficient CO2RR catalysts.
A Cu‐porphyrin‐based large‐scale, ultrathin nanosheet with numerous hydrogen‐bonding networks was developed for the highly selective electroreduction of CO2 to CH4 under neutral conditions. This catalyst exhibits a superior FECH4
of 70 % with a high current density (−183.0 mA cm−2) at −1.6 V under rarely reported neutral conditions and maintains FECH4
>51 % over a wide potential range (−1.5 to −1.7 V) in a flow cell.
Background
Coronavirus disease 2019 (COVID‐19) caused by severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection has been widely spread. We aim to investigate the clinical ...characteristic and allergy status of patients infected with SARS‐CoV‐2.
Methods
Electronic medical records including demographics, clinical manifestation, comorbidities, laboratory data, and radiological materials of 140 hospitalized COVID‐19 patients, with confirmed result of SARS‐CoV‐2 viral infection, were extracted and analyzed.
Results
An approximately 1:1 ratio of male (50.7%) and female COVID‐19 patients was found, with an overall median age of 57.0 years. All patients were community‐acquired cases. Fever (91.7%), cough (75.0%), fatigue (75.0%), and gastrointestinal symptoms (39.6%) were the most common clinical manifestations, whereas hypertension (30.0%) and diabetes mellitus (12.1%) were the most common comorbidities. Drug hypersensitivity (11.4%) and urticaria (1.4%) were self‐reported by several patients. Asthma or other allergic diseases were not reported by any of the patients. Chronic obstructive pulmonary disease (COPD, 1.4%) patients and current smokers (1.4%) were rare. Bilateral ground‐glass or patchy opacity (89.6%) was the most common sign of radiological finding. Lymphopenia (75.4%) and eosinopenia (52.9%) were observed in most patients. Blood eosinophil counts correlate positively with lymphocyte counts in severe (r = .486, P < .001) and nonsevere (r = .469, P < .001) patients after hospital admission. Significantly higher levels of D‐dimer, C‐reactive protein, and procalcitonin were associated with severe patients compared to nonsevere patients (all P < .001).
Conclusion
Detailed clinical investigation of 140 hospitalized COVID‐19 cases suggests eosinopenia together with lymphopenia may be a potential indicator for diagnosis. Allergic diseases, asthma, and COPD are not risk factors for SARS‐CoV‐2 infection. Older age, high number of comorbidities, and more prominent laboratory abnormalities were associated with severe patients.
Decreased eosinophil count, which was positively correlated with lymphocyte counts, may be a potential biological indicator for diagnosing
COVID‐19 patients. Low prevalence of allergic diseases, COPD and patients with smoking history indicated they may not be the predisposing
factors of COVID‐19. Elder age, high number of comorbidities and more prominent laboratory abnormalities were associated with severe
patientsz.
Conventional neural architecture search (NAS) approaches are usually based on reinforcement learning or evolutionary strategy, which take more than 1000 GPU hours to find a good model on CIFAR-10. We ...propose an efficient NAS approach, which learns the searching approach by gradient descent. Our approach represents the search space as a directed acyclic graph (DAG). This DAG contains thousands of sub-graphs, each of which indicates a kind of neural architecture. To avoid traversing all the possibilities of the sub-graphs, we develop a differentiable sampler over the DAG. This sampler is learnable and optimized by the validation loss after training the sampled architecture. In this way, our approach can be trained in an end-to-end fashion by gradient descent, named Gradient-based search using Differentiable Architecture Sampler (GDAS). In experiments, we can finish one searching procedure in four GPU hours on CIFAR-10, and the discovered model obtains a test error of 2.82% with only 2.5M parameters, which is on par with the state-of-the-art.
PDF
