Search

Background A significant roadblock in surgical education research has been the inability to compare trainee performance to the outcomes of those surgeons after they enter independent practice. We describe the feasibility of an innovative method to link trainee performance data with patient outcomes. Methods We extracted surgeon NPI numbers from Medicare claims data for common general surgery procedures between 2007 and 2017. Next, American Board of Surgery (ABS) trainee performance data was cross-referenced with additional resources to supplement NPI data. The patient and trainee datasets were linked using NPI number and a linkage rate was calculated. Results We identified 12,952 unique surgeons in the Medicare file. Medicare surgeons were matched with ABS records by NPI number, with 96.2% (n = 12,460) of surgeons linked successfully. Conclusions We demonstrated a novel process to link patient outcomes to trainee performance. This innovation can enable future research investigating the relationship between surgical trainee performance and patient outcomes in independent practice.

Objective Surgical trainees are subject to pressure from variety of stakeholders to secure board certification from the American Board of Surgery (ABS). To meet these expectations, trainees must pass a written qualifying exam (QE) and an oral certifying exam (CE) within 7 years of completing general surgery residency. Board certification outcomes for candidates who fail either the QE or CE examination are not well characterized, but this information could help candidates, policymakers, and other stakeholders make informed decisions about how to respond to examination failure. Methods We retrospectively examined ABS records for all surgeons who completed general surgery residency from 2000 to 2013 and attempted general surgery board certification. Results Among 14,483 surgeons who attempted general surgery certification, 13,566 (94%) passed both the QE and CE within the 7-year certification window. Of those who did ultimately obtain certification, 97% passed the QE within 2 attempts and 97% passed the CE within 2 attempts. For those who failed either the QE or the CE twice, 67% ultimately obtained certification. Conclusions Most surgeons who obtained ABS general surgery board certification did so within 2 attempts at each board examination. Candidates who fail either examination twice are less likely to achieve board certification.

Learning the conjunction of multiple visual concepts shows practical significance in various real world applications (e.g. multi-attribute image retrieval and visual relationship detection). In this paper, we propose Concept Conjunction Recurrent Neural Network (C2RNN) to tackle this problem. With our model, visual concepts involved in a conjunction are mapped into the hidden units and combined in a recurrent way to generate the representation of the concept conjunction, which is then used to compute a concept conjunction classifier as the output. We also present an order invariant version of the proposed method based on attention mechanism to learn the tasks without pre-defined concept order. To tackle concept conjunction learning from multiple semantic domains, we introduce a multiplicative framework to learn the joint representation. Experimental results on multi-attribute image retrieval and visual relationship detection show that our method achieves significantly better performance than other related methods on various datasets.

In this work, we study the problem of deep representation learning on a large face dataset with long-tail distribution. Training convolutional neural networks on such dataset with conventional strategy suffers from imbalance problem which results in biased classification boundary, and the few-shot classes lying in tail parts further make the model prone to overfitting. Aiming to learn more discriminative CNN model from long-tail data, we propose a novel aggregate-and-disperse training schema. Firstly, our proposed method aggregates similar classes in tail part to avoid imbalance problem. Based on the aggregated super classes and those original head classes, a model is pre-trained to capture accurate discrimination in head classes as well as coarse discrinimation in tail classes. Secondly, we selectively disperses those aggregated super classes to learn precise inter-class variations and refine the representation for better generalization. We perform extensive experiments on MS-Celeb-1M, BLUFR and MegaFace. Compared with baselines and existing methods, our method achieves better performance of face recognition, demonstrating its effectiveness of handling long-tail distribution.

CaO-Al2O3-SiO2(CAS) glass ceramics were designed and prepared using a melt-quench approach. The coefficient of the thermal expansion (CTE) of the synthesized CAS (4.12 × 10−6 K−1) matched perfectly with that of the SiC ceramic (4.01 × 10−6 K−1). Thermal analysis of the CAS was conducted. Then the joining of the SiC ceramics by the CAS glass ceramics under various process parameters were conducted. The bonding temperature affects the fluidity of the CAS glass and the oxidation of the SiC substrate. The holding duration decides the infiltration of the CAS glass into the SiC substrate. The optimal bonding parameter is 1400 ℃/10 min and the corresponding highest shear strength of the SiC/CAS/SiC bonded joints in average was 56 MPa. Fracture observation was also conducted to help analyze the relationship between the interfacial microstructure and the joint strength. Finally, the formation mechanism of the SiC/CAS/SiC bonded joints was proposed.

Objective The profession of surgery is entering a new era of “big data,” where analyses of longitudinal trainee assessment data will be used to inform ongoing efforts to improve surgical education. Given the high-stakes implications of these types of analyses, researchers must define the conditions under which estimates derived from these large datasets remain valid. With this study, we determine the number of assessments of residents’ performances needed to reliably assess the difficulty of “Core” surgical procedures. Design Using the SIMPL smartphone application from the Procedural Learning and Safety Collaborative, 402 attending surgeons directly observed and provided workplace-based assessments for 488 categorical residents after 5259 performances of 87 Core surgical procedures performed at 14 institutions. We used these faculty ratings to construct a linear mixed model with resident performance as the outcome variable and multiple predictors including, most significantly, the operative procedure as a random effect. We interpreted the variance in performance ratings attributable to the procedure, after controlling for other variables, as the “difficulty” of performing the procedure. We conducted a generalizability analysis and decision study to estimate the number of SIMPL performance ratings needed to reliably estimate the difficulty of a typical Core procedure. Results Twenty-four faculty ratings of resident operative performance were necessary to reliably estimate the difficulty of a typical Core surgical procedure (mean dependability coefficient 0.80, 95% confidence interval 0.73-0.87). Conclusions At least 24 operative performance ratings are required to reliably estimate the difficulty of a typical Core surgical procedure. Future research using performance ratings to establish procedure difficulty should include adequate numbers of ratings given the high-stakes implications of those results for curriculum design and policy.

Monitoring of fuel element failure in a nuclear reactor is an important process to ensure safe operation of the nuclear reactor. It is feasible to detect fuel element rupture by analysing the radionuclides of fission products and activation products in the primary coolant. To identify the radionuclides in the primary coolant, the most convenient method to measure their activity concentrations is γ-ray spectrometry. A γ-ray spectral analysis method for real-time detection of fuel element failure is proposed in this paper. The method aims to implement a Gaussian smoothing processing for spectral data, extract peaks of spectral data through mathematical morphological transformation, determine left and right peak boundaries with a gradient descent algorithm and finally estimate net peak area with an image method. The method can rapidly recognize a designated key radionuclide from an obtained spectrum and extract peak information to meet the demands of real-time monitoring of fuel element failure in pressurized water reactors.

Hereditary angioedema (HAE) is a rare and potentially life-threatening disease that affects an estimated 1 in 50,000 individuals worldwide. Berotralstat (BCX7353) is the only small molecule approved by the US Food and Drug Administration (FDA) for the prophylactic treatment of HAE attacks in patients 12 years and older. During the discovery of BCX7353, we also identified a novel series of small molecules containing a quaternary carbon as potent and orally bioavailable Plasma Kallikrein (PKal) inhibitors. Lead compound was identified as a potent inhibitor following a detailed lead optimization process that balanced the lipophilic efficiency (LipE) and pharmacokinetic (PK) profile.

Heavy metal contamination in agricultural soil and immobilization remediation have generated widespread concern in all areas of society. Sulfhydryl-functionalized materials as emerging amendments exhibit application potential, but the dose-effect relationship and immobilization mechanism are poorly understood. To understand the relationship between the immobilization effect and total sulfhydryl content, sulfhydryl-grafted palygorskite (SGP) with three sulfhydryl loading rates (0.88 mmol/g, 1.83 mmol/g, and 2.77 mmol/g) was prepared and characterized in the current study. The Cd immobilization efficiency and the dose-effect relationship were investigated via sorption in solutions, soil incubation, and field-scale wheat cultivation.

Inspired by the concept of memory mechanism and predictive coding from the cognitive neuroscience, this paper presents a deep memory and prediction neural network (DMPNet) for video prediction. Correspondingly, memory and error propagation units are designed in DMPNet to capture the previous spatial-temporal information and compute current predictive error which is forwarded to the prediction unit for correcting the subsequent video prediction. Subsequently, prediction unit takes the information stored in memory unit and predictive error of previous frame as input to predict the next frame. We evaluate our method on two public real-world datasets and demonstrate that the proposed DMPNet outperforms some state-of-the-art methods quantitatively and qualitatively.

The Risk Analysis Index (RAI) for frailty is a rapid survey for comorbidities and performance status, which predicts mortality after general surgery. We aimed to validate the RAI in predicting outcomes after hepatopancreatobiliary surgery.Associations of RAI, determined in 162 patients prior to undergoing hepatopancreatobiliary surgery, with prospectively collected 30-day post-operative outcomes were analyzed with multivariate logistic and linear regression.Patients (age 62 ± 14, 51% female) had a median RAI of 7, range 0-25. With every unit increase in RAI, length of stay increased by 5% (95% CI: 2-7%), odds of ICU admission increased by 10% (0-20%), ICU length of stay increased by 21% (9-34%), and odds of discharge to a nursing facility increased by 8% (0-17%) (all P 0.05). Particularly in patients who suffered a first post-operative complication, RAI was associated with additional complications (1.6 unit increase in Comprehensive Complication Index per unit increase in RAI, P = 0.002). In a direct comparison in a subset of 74 patients, RAI and the ACS-NSQIP Risk Calculator performed comparably in predicting outcomes.While RAI and ACS-NSQIP Risk Calculator comparatively predicted short-term outcomes after HPB surgery, RAI has been specifically designed to identify frail patients who can potentially benefit from preoperative prehabilitation interventions.

Attributes are widely used in different vision tasks. However, existing attribute resources are quite limited and most of them are not in large scale. Current attribute annotation process is generally done by human, which is expensive and time-consuming. In this paper, we propose a novel framework to perform effective attribute annotations. Based on the common knowledge that attributes can be shared among different classes, we leverage the benefits of transfer learning and active learning together to transfer knowledge from some existing small attribute databases to large-scale target databases. In order to learn more robust attribute models, attribute relationships are incorporated to assist the learning process. Using the proposed framework, we conduct extensive experiments on two large-scale image databases, i.e. ImageNet and SUN Attribute, where high quality automatic attribute annotations are obtained.

In indirectly driven inertial confinement fusion, a high-purity germanium detector is widely used in the calibration of various X-ray detectors as a quantitative spectrum detector for the irradiating X-ray. However, the intrinsic peak efficiency of the high-purity germanium detector, which is the foundation of its applications, has not been investigated in depth. In this work, an optimized data-processing method is developed to explore the intrinsic peak efficiency of the high-purity germanium detector, based on calibration experiments. The intrinsic peak efficiency of the detector is absolutely and accurately calibrated and simulated for X-ray energy in the range of 5.48–302.85 keV. The experimental and simulated results, which agree well, are analyzed and explained in detail. The efficiency curve obtained from the data reveals the rise, flat, and fall characters. In addition, the impact of the absorption edge, which is located at 11.10 keV, on the efficiency curve was found and explained in terms of the escape of the excited K α lines. This work will improve the calibration efficiency of various X-ray detectors based on the high-purity germanium detector, and help in advancing the quantitative understanding of the detection results in inertial confinement fusion.

Learning an appropriate distance metric is a crucial problem in pattern recognition. To confront with the scalability issue of massive data, hamming distance on binary codes is advocated since it permits exact sub-linear kNN search and meanwhile shares the advantage of efficient storage. In this paper, we study hamming metric learning in the context of multimodal data for cross-view similarity search. We present a new method called Parametric Local Multiview Hamming metric (PLMH), which learns multiview metric based on a set of local hash functions to locally adapt to the data structure of each modality. To balance locality and computational efficiency, the hash projection matrix of each instance is parameterized, with guaranteed approximation error bound, as a linear combination of basis hash projections associated with a small set of anchor points. The weak-supervisory information (side information) provided by pairwise and triplet constraints are incorporated in a coherent way to achieve semantically effective hash codes. A local optimal conjugate gradient algorithm with orthogonal rotations is designed to learn the hash functions for each bit, and the overall hash codes are learned in a sequential manner to progressively minimize the bias. Experimental evaluations on cross-media retrieval tasks demonstrate that PLMH performs competitively against the state-of-the-art methods.

Due to the environmental differences, many face anti-spoofing methods fail to generalize to unseen scenarios. In light of this, we propose a unified unsupervised and semi-supervised domain adaptation network (USDAN) for cross-scenario face anti-spoofing, aiming at minimizing the distribution discrepancy between the source and the target domains. Specifically, two modules, i.e., marginal distribution alignment module (MDA) and conditional distribution alignment module (CDA), are designed to seek a domain-invariant feature space via adversarial learning and make the features of the same class compact, respectively. By adding/removing the CDA module, the network can be easily switched for semi-supervised/unsupervised setting, in which sense our method is named with “unified”. Moreover, the adaptive cross-entropy loss and normalization techniques are further incorporated to improve the generalization. Extensive experimental results show that the proposed USDAN outperforms state-of-the-art methods on several public datasets.

In this paper, we address one specific video retrieval problem in terms of human face. Given one query in forms of either a frame or a sequence from a person, we search the database and return the most relevant face videos, i.e., ones have the same class label with the query. Such problem is very challenging due to the large intra-class variations and the high request on the efficiency of video representations in terms of both time and space. To handle such challenges, this paper proposes a novel Deep Video Code (DVC) method which encodes video faces into compact binary codes. Specifically, we devise an end-to-end convolutional neural network (CNN) framework that takes face videos as training inputs, models each of them as a unified representation by temporal feature pooling operation, and finally projects the high-dimensional representations of both frames and videos into Hamming space to generate binary codes. In such Hamming space, distance of dissimilar pairs is larger than that of similar pairs by a margin. To this end, a novel bounded triplet hashing loss is elaborately designed, which takes all dissimilar pairs into consideration for each anchor point in a mini-batch, and the optimization of the loss function is smoother and more stable. Extensive experiments on challenging video face databases and general image/video datasets with comparison to the state-of-the-arts verify the effectiveness of our method in different kinds of retrieval scenarios.

Multi-view face detection in open environment is a challenging task due to diverse variations of face appearances and shapes. Most multi-view face detectors depend on multiple models and organize them in parallel, pyramid or tree structure, which compromise between the accuracy and time-cost. Aiming at a more favorable multi-view face detector, we propose a novel funnel-structured cascade (FuSt) detection framework. In a coarse-to-fine flavor, our FuSt consists of, from top to bottom, 1) multiple view-specific fast LAB cascade for extremely quick face proposal, 2) multiple coarse MLP cascade for further candidate window verification, and 3) a unified fine MLP cascade with shape-indexed features for accurate face detection. Compared with other structures, on the one hand, the proposed one uses multiple computationally efficient distributed classifiers to propose a small number of candidate windows but with a high recall of multi-view faces. On the other hand, by using a unified MLP cascade to examine proposals of all views in a centralized style, it provides a favorable solution for multi-view face detection with high accuracy and low time-cost. Besides, the FuSt detector is alignment-aware and performs a coarse facial part prediction which is beneficial for subsequent face alignment. Extensive experiments on two challenging datasets, FDDB and AFW, demonstrate the effectiveness of our FuSt detector in both accuracy and speed., Comment: Submitted to Neurocomputing (under review). An adapted open source implementation can be found at https://github.com/seetaface/SeetaFaceEngine/tree/master/FaceDetection

Video-based spontaneous expression recognition is a challenging task due to the large inter-personal variations of both the expressing manners and the executing rates for the same expression category. One of the key is to explore robust representation method which can effectively capture the facial variations as well as alleviate the influence of personalities. In this paper, we propose to learn a kind of typical patterns that can be commonly shared by different subjects when performing expressions, namely "prototypes". Specifically, we first apply a statistical model (i.e. linear subspace) on facial regions to generate the specific expression patterns for each video. Then a clustering algorithm is employed on all these expression patterns and the cluster means are regarded as the "prototypes". Accordingly, we further design "simile" features to measure the similarities of personal specific patterns to our learned "prototypes". Both techniques are conducted on Grassmann manifold, which can enrich the feature encoding manners and better reveal the data structure by introducing intrinsic geodesics. Extensive experiments are conducted on both posed and spontaneous expression databases. All results show that our method outperforms the state-of-the-art and also possesses good transferable ability under cross-database scenario.

Rechargeable lithium (Li) metal batteries with conventional LiPF6-carbonate electrolytes have been reported to fail quickly at charging current densities of about 1.0 mA cm−2 and above. In this work, we demonstrate the rapid charging capability of Li||LiNi0.8Co0.15Al0.05O2 (NCA) cells can be enabled by a dual-salt electrolyte of lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) and lithium bis(oxalato)borate (LiBOB) in a carbonate solvent mixture. The cells using the LiTFSI-LiBOB dual-salt electrolyte significantly outperform those using the LiPF6 electrolyte at high charging current densities. At the charging current density of 1.50 mA cm−2, the Li||NCA cells with the dual-salt electrolyte can still deliver a discharge capacity of 131 mAh g−1 and a capacity retention of 80% after 100 cycles. The Li||NCA cells with the LiPF6 electrolyte start to show fast capacity fading after the 30th cycle and only exhibit a low capacity of 25 mAh g−1 and a low retention of 15% after 100 cycles. The reasons for the good chargeability and cycling stability of the cells using the LiTFSI-LiBOB dual-salt electrolyte can be attributed to the good film-formation ability of the electrolyte on the Li metal anode and the highly conductive nature of the sulfur-rich interphase layer.

The analysis of fuel rod failure character is the key to a real-time detection system for fuel rod failure in a pressurized water reactor (PWR). We introduce an analysis code for fuel rod failure character in PWRs that allows online analysis and offline usage. The fuel rod failure data from some nuclear reactors were used to verify the analysis code. The results show that the number of failed fuel rods calculated by the analysis code based on the isotopes of Kr and Xe in the fission gases fits very well with the actual number of failed fuel rods. The real-time detection system for fuel rod failure in PWRs and the analysis code for fuel rod failure character are of great significance for the safe operation of nuclear reactors.

Unconstrained face recognition still remains a challenging task due to various factors such as pose, expression, illumination, partial occlusion, etc. In particular, the most significant appearance variations are stemmed from poses which leads to severe performance degeneration. In this paper, we propose a novel Deformable Face Net (DFN) to handle the pose variations for face recognition. The deformable convolution module attempts to simultaneously learn face recognition oriented alignment and identity-preserving feature extraction. The displacement consistency loss (DCL) is proposed as a regularization term to enforce the learnt displacement fields for aligning faces to be locally consistent both in the orientation and amplitude since faces possess strong structure. Moreover, the identity consistency loss (ICL) and the pose-triplet loss (PTL) are designed to minimize the intra-class feature variation caused by different poses and maximize the inter-class feature distance under the same poses. The proposed DFN can effectively handle pose invariant face recognition (PIFR). Extensive experiments show that the proposed DFN outperforms the state-of-the-art methods, especially on the datasets with large poses.

Cesium ion (Cs+) has been reported to be an effective electrolyte additive to suppress Li dendrite growth which prevents the application of lithium (Li) metal as an anode for rechargeable Li batteries. In this work, we investigated the effect of Cs+ additive on Li depositions using quantitative in situ 7Li and 133Cs nuclear magnetic resonance (NMR) with planar symmetric Li cells. It's found that the addition of Cs+ can significantly enhance both the formation of well aligned Li nanorods and reversibility of the Li electrode. In situ 133Cs NMR directly confirms that Cs+ migrates to Li electrode to form a positively charged electrostatic shield during the charging process. Much more electrochemical “active” Li was found in Li films deposited with Cs+ additive, while more electrochemical “dead” and thicker Li rods were identified in Li films deposited without Cs+. Combining the in situ and the previous ex-situ results, a Li deposition model has been proposed to explain these observations.

In this paper, we propose a method for robust Sign Language Recognition from RGB-D data. A Sparse Observation (SO) description is proposed to character each sign in terms of the typical hand postures. Concretely speaking, the SOs are generated by considering the typical posture fragments, where hand motions are relatively slow and hand shapes are stable. Thus the matching between two sign words is converted to measure the similarity computing between two aligned SO sequences. The alignment is formulated as a variation of Stable Marriage Problem (SMP). The classical "propose-engage" idea is extended to get the order preserving matched SO pairs. In the training stage, the multiple instances from one sign are fused to generate single SO template. In the recognition stage, SOs of each probe sign "propose" to SOs of the templates for the purpose of reasonable similarity computing. To further speed up the SO alignment, hand posture relationship map is constructed as a strong prior to generate the distinguished low-dimensional feature of SO. Moreover, to get much better performance, the motion trajectory feature is integrated. Experiments on two large datasets and an extra Chalearn Multi-modal Gesture Dataset demonstrate that our algorithm has much higher accuracy with only 1/10 time cost compared with the HMM and DTW based methods.

The practical application of lithium (Li) metal anode in rechargeable Li batteries is hindered by both the growth of Li dendrites and the low Coulombic efficiency (CE) during repeated charge/discharge cycles. Recently, we have discovered that CsPF 6 as an electrolyte additive can significantly suppress Li dendrite growth and lead to highly compacted and well aligned Li nanorod structures during Li deposition on copper substrates. In this paper, the effect of CsPF 6 additive on the performance of rechargeable Li metal batteries with lithium iron phosphate (LFP) cathode is further studied. Li|LFP coin cells with CsPF 6 additive in electrolytes show well protected Li anode surface, decreased resistance, enhanced rate capability and extended cycling stability. In Li|LFP cells, the electrolyte with CsPF 6 additive shows excellent long-term cycling stability (at least 500 cycles) at a charge current density of 0.5 mA cm −2 without internal short circuit. At high charge current densities, the effect of CsPF 6 additive becomes less significant. Future work needs to be done to protect Li metal anode, especially at high charge current densities and for long cycle life.

Face recognition on large-scale video in the wild is becoming increasingly important due to the ubiquity of video data captured by surveillance cameras, handheld devices, Internet uploads, and other sources. By treating each video as one image set, set-based methods recently have made great success in the field of video-based face recognition. In the wild world, videos often contain extremely complex data variations and thus pose a big challenge of set modeling for set-based methods. In this paper, we propose a novel Hybrid Euclidean-and-Riemannian Metric Learning (HERML) method to fuse multiple statistics of image set. Specifically, we represent each image set simultaneously by mean, covariance matrix and Gaussian distribution, which generally complement each other in the aspect of set modeling. However, it is not trivial to fuse them since mean, covariance matrix and Gaussian model typically lie in multiple heterogeneous spaces equipped with Euclidean or Riemannian metric. Therefore, we first implicitly map the original statistics into high dimensional Hilbert spaces by exploiting Euclidean and Riemannian kernels. With a LogDet divergence based objective function, the hybrid kernels are then fused by our hybrid metric learning framework, which can efficiently perform the fusing procedure on large-scale videos. The proposed method is evaluated on four public and challenging large-scale video face datasets. Extensive experimental results demonstrate that our method has a clear superiority over the state-of-the-art set-based methods for large-scale video-based face recognition. HighlightsRepresent image set by mean, covariance and Gaussian for discriminant information.Heterogeneous Euclidean and Riemannian kernels are exploited and fused clearly.Clear superiority over state-of-the-art set-based methods is achieved in testing.

Most existing technologies for facial expression recognition utilize off-the-shelf feature extraction methods for classification. In this paper, aiming at learning better features specific for expression representation, we propose to construct a deep architecture, AU-inspired Deep Networks (AUDN), inspired by the psychological theory that expressions can be decomposed into multiple facial Action Units (AUs). To fully exploit this inspiration but avoid detecting AUs, we propose to automatically learn: (1) informative local appearance variation; (2) optimal way to combining local variation and (3) high level representation for final expression recognition. Accordingly, the proposed AUDN is composed of three sequential modules. Firstly, we build a convolutional layer and a max-pooling layer to learn the Micro-Action-Pattern (MAP) representation, which can explicitly depict local appearance variations caused by facial expressions. Secondly, feature grouping is applied to simulate larger receptive fields by combining correlated MAPs adaptively, aiming to generate more abstract mid-level semantics. Finally, a multi-layer learning process is employed in each receptive field respectively to construct group-wise sub-networks for higher-level representations. Experiments on three expression databases CK+, MMI and SFEW demonstrate that, by simply applying linear classifiers on the learned features, our method can achieve state-of-the-art results on all the databases, which validates the effectiveness of AUDN in both lab-controlled and wild environments.

Canonical Correlation Analysis (CCA) is one of the most popular statistical methods to capture the correlations between two variables. However, it has limitations as a linear and global algorithm. Although some variants have been proposed to overcome the limitations, neither of them achieves locality and nonlinearity at the same time. In this paper, we propose a novel algorithm called Instance-Specific Canonical Correlation Analysis (ISCCA), which approximates the nonlinear data by computing the instance-specific projections along the smooth curve of the manifold. First, we propose a least squares solution for CCA which will set the stage for the proposed method. Second, based on the framework of least squares regression, CCA is extended to the instance-specific case which obtains a set of locally linear smooth but globally nonlinear transformations. Third, ISCCA can be extended to semi-supervised setting by exploiting the unlabeled data to further improve the performance. The optimization problem is proved to be convex and could be solved efficiently by alternating optimization. And the globally optimal solutions could be achieved with theoretical guarantee. Moreover, for large scale applications, iterative conjugate gradient algorithm can be used to speed up the computation procedure. Experimental results demonstrate the effectiveness of our proposed method. HighlightsWe model locality and nonlinearity jointly for multi-view correlation learning.Instance-specific projections are computed along the smooth curve of the manifold.The convex objective functions are solved efficiently with a global optimal solution.

Solar domestic hot water system is widely used and developing fast in recent years in China. However many problems occur at the same time, for example more energy consumption by circulation pump, water reheating, long investment payback period, and etc. Through analyzing the field-testing data of projects and investigating of different residential consumers, it was found that compared to actual hot water consumption the solar domestic hot water systems were generally designed too large in capacity, which means the designed hot water demand is much greater than actual user consumption. This study compared different specifications and recommended design parameters value of hot water related standards, and analyzed the calculation methodology and design parameter ofhot water quota. Finally problems in the system design are summarized and suggestions are proposed for designers and different stakeholders.

Solar fraction ratio is a key index and reference of solar hot water system design, andis also a key factor to evaluate solar hot water system according to Evaluation Standard for Application of Renewable Energy in Buildings in China. By analyzing relevant inspection data of actual projects, it was found that using solar fraction ratio to evaluate the actual running systems has certain limitation, which cannot reasonably reflect the actual supplementation of conventional energy, especially with the residential buildings applying central solar hot water system. Based on the total energy consumption control concept raised by government during the Twelfth Five-Year Plan period, the actual supplementation level of conventional energy should be used as a factor to evaluate solar hot water system. This study will analyze the limitation of solar fraction ratio in design and evaluation, and propose corresponding ideas of solution as references for relevant design and evaluation professionals.

A novel Sparsely Encoded Local Descriptor (SELD) is proposed for face verification. Different from traditional hard or soft quantization methods, we exploit linear regression (LR) model with sparsity and non-negativity constraints to extract more discriminative features (i.e. sparse codes) from local image patches sampled pixel-wisely. Sum-pooling is then imposed to integrate all the sparse codes within each block partitioned from the whole face image. Whitened Principal Component Analysis (WPCA) is finally used to suppress noises and reduce the dimensionality of the pooled features, which thus results in the so-called SELD. To validate the proposed method, comprehensive experiments are conducted on face verification task to compare SELD with the existing related methods in terms of three variable component modules: K-means or K-SVD for dictionary learning, hard/soft assignment or regression model for encoding, as well as sum-pooling or max-pooling for pooling. Experimental results show that our method achieves a competitive accuracy compared with the state-of-the-art methods on the challenging Labeled Faces in the Wild (LFW) database.

Hair segmentation is challenging due to the diverse appearance, irregular region boundary and the influence of complex background. To deal with this problem, we propose a novel data-driven method, named Isomorphic Manifold Inference (IMI). The IMI method assumes the coarse probability map and the binary segmentation map as a couple of isomorphic manifolds and tries to learn hair specific priors from manually labeled training images. For an input image, firstly, the method calculates a coarse probability map. Then it exploits regression techniques to obtain the relationship between the coarse probability map of the test image and those of training images. Finally, this relationship, i.e., a coefficient set, is transferred to the binary segmentation maps and a soft segmentation of the test image will be achieved by a linear combination of those binary maps. Further, we employ this soft segmentation as a shape cue and integrate it with color and texture cues into a unified segmentation framework. A better segmentation is achieved by the Graph Cuts optimization. Extensive experiments are conducted to validate effectiveness of the IMI method, compare contributions of different cues and investigate the generalization of IMI method. The results strongly encourage our method.

Due to the imbalance of heating and cooling demand, the soil temperature decreases after several years' operation for soil source heat pump system, and consequently the coefficient of performance (COP) decreases. Soil heat charging with solar energy was proposed for solar-soil source heat pump compound system. Many studies have been carried out to investigate the heat and moisture transfer in soil for soil heat charging, but they failed to provide a clear description and explanation of the coupled and mutual effect between heat and moisture transfer. This paper presents an experimental study on the heat and moisture transfer in soil heat charging. The simplified testing rig was constructed to investigate the coupled and mutual effect of one-dimension horizontal heat and moisture transfer in soil heat charging. The results show that, peak volumetric water content (VWC) occurs firstly at the position near heat source and moves towards the other cold end of the tested soil pillar, but it is limited in an area near heat source. The time consumption and value for peak VWC increase with the increasing heat source temperature at the same initial VWC and increase with the increasing initial VWC at the same heat source temperature as well. The heat source temperature is a main factor affecting the peak soil temperature and peak VWC, while the initial VWC has little impact on them.

Video-based Face Recognition (VFR) can be converted into the problem of measuring the similarity of two image sets, where the examples from a video clip construct one image set. In this paper, we consider face images from each clip as an ensemble and formulate VFR into the Joint Sparse Representation (JSR) problem. In JSR, to adaptively learn the sparse representation of a probe clip, we simultaneously consider the class-level and atom-level sparsity, where the former structurizes the enrolled clips using the structured sparse regularizer (i.e., L 2 , 1 -norm) and the latter seeks for a few related examples using the sparse regularizer (i.e., L 1 - norm ). Besides, we also consider to pre-train a compacted dictionary to accelerate the algorithm, and impose the non-negativity constraint on the recovered coefficients to encourage positive correlations of the representation. The classification is ruled in favor of the class that has the lowest accumulated reconstruction error. We conduct extensive experiments on three real-world databases: Honda, MoBo and YouTube Celebrities (YTC). The results demonstrate that our method is more competitive than those state-of-the-art VFR methods. HighlightsPropose a reconstruction-based method for video-based face recognition, where all images from a probe video clip are jointly recovered.Use two sparse constraints to make representation more credible.Develop an efficient optimization algorithm.Get a more competitive performance on the challenging dataset YTC.

Lithium ion batteries (LIBs) are attracting intensive interests worldwide because of their potential applications in transportation electrification and utility grid. The intercalation compounds used in LIBs electrochemically react with Li + ions via single or multiple phase transitions depending on the nature of the material structure as well as the synthesis and operating conditions. For LiNi 0.5 Mn 1.5 O 4 high voltage spinel, a promising candidate positive electrode material for LIBs, there are three spinel-structured phases sequentially appeared through two successive two-phase reactions during the delithiation/lithiation processes. Here we demonstrate, experimentally and theoretically, that through elemental substitution, the solid solution ranges for both the first and second phases are significantly extended during the electrochemical charge–discharge process. This type of structural changes with more solid solution regions facilitate fast Li + diffusion by reducing the number of phase boundaries that Li + ions have to overcome and resulted in less shrinkage of the unit cells at the end of charge process. This work unravels the fundamental interactions between structural and electrochemical properties by using spinel as the platform, which may be widely adopted to explain or tailor the properties of materials for energy storage and conversion.

In this paper, we present a gesture recognition approach to enable real-time manipulating projection content through detecting and recognizing speakers gestures from the depth maps captured by a depth sensor. To overcome the limited measurement accuracy of depth sensor, a robust background subtraction method is proposed for effective human body segmentation and a distance map is adopted to detect human hands. Potential Active Region (PAR) is utilized to ensure the generation of valid hand trajectory to avoid extra computational cost on the recognition of meaningless gestures and three different detection modes are designed for complexity reduction. The detected hand trajectory is temporally segmented into a series of movements, which are represented as Motion History Images. A set-based soft discriminative model is proposed to recognize gestures from these movements. The proposed approach is evaluated on our dataset and performs efficiently and robustly with 90% accuracy.

Face recognition from Single Sample per Person (SSPP) is extremely challenging because only one sample is available for each person. While many discriminant analysis methods, such as Fisherfaces and its numerous variants, have achieved great success in face recognition, these methods cannot work in this scenario, because more than one sample per person are needed to calculate the within-class scatter matrix. To address this problem, we propose Adaptive Discriminant Analysis (ADA) in which the within-class scatter matrix of each enrolled subject is inferred using his/her single sample, by leveraging a generic set with multiple samples per person. Our method is motivated from the assumption that subjects who look alike to each other generally share similar within-class variations. In ADA, a limited number of neighbors for each single sample are first determined from the generic set by using kNN regression or Lasso regression. Then, the within-class scatter matrix of this single sample is inferred as the weighted average of the within-class scatter matrices of these neighbors based on the arithmetic mean or Riemannian mean. Finally, the optimal ADA projection directions can be computed analytically by using the inferred within-class scatter matrices and the actual between-class scatter matrix. The proposed method is evaluated on three databases including FERET database, FRGC database and a large real-world passport-like face database. The extensive results demonstrate the effectiveness of our ADA when compared with the existing solutions to the SSPP problem.

The commercial Li4Ti5O12 (LTO) is successfully modified by AlF3 via a low temperature process. After being calcined at 400 °C for 5 h, AlF3 reacts with LTO to form a composite material which mainly consists of Al3+ and F− co-doped LTO with small amounts of anatase TiO2. Al3+ and F− co-doped LTO demonstrates ultrahigh rate capability comparing to the pristine LTO. Since the amount of the byproduct TiO2 is relatively small, the modified LTO electrodes retain the main voltage characteristics of LTO with a minor feature similar to those of anatase TiO2. The doped LTO anodes deliver slightly higher discharge capacity and maintain the excellent long-term cycling stability when compared to the pristine LTO anode. Therefore, Al3+ and F− co-doped LTO composite material synthesized at low temperature is an excellent anode for stable and ultra-high power lithium-ion batteries.

Illumination preprocessing is an effective and efficient approach in handling lighting variations for face recognition. Despite much attention to face illumination preprocessing, there is seldom systemic comparative study on existing approaches that presents fascinating insights and conclusions in how to design better illumination preprocessing methods. To fill this vacancy, we provide a comparative study of 12 representative illumination preprocessing methods (HE, LT, GIC, DGD, LoG, SSR, GHP, SQI, LDCT, LTV, LN and TT) from two novel perspectives: (1) localization for holistic approach and (2) integration of large-scale and small-scale feature bands. Experiments on public face databases (YaleBExt, CMU-PIE, CAS-PEAL and FRGC V2.0) with illumination variations suggest that localization for holistic illumination preprocessing methods (HE, GIC, LTV and TT) further improves the performance. Integration of large-scale and small-scale feature bands for reflectance field estimation based illumination preprocessing approaches (SSR, GHP, SQI, LDCT, LTV and TT) is also found helpful for illumination-insensitive face recognition.

The stabilities of different conductive carbon additives have been systematically investigated in high voltage lithium ion batteries. It is found that the higher surface area of conductive additives leads to more parasitic reactions initiating from different onset voltages. A closer inspection reveals that for the low surface area carbon such as Super P, PF 6 − anions reversibly intercalate into carbon structure at around 4.7 V. For high surface area carbons, in addition to the electrolyte decomposition, the oxidation of functional groups at high voltage further increases the irreversible capacity and Li+ ion consumption. Coulombic efficiency, irreversible capacity and cycling stability observed by using different carbon additives are correlated with their structure and surface chemistry, thus providing information for predictive selection of carbon additives in different energy storage systems.

It is well known that LATP (Li1+x+yAlxTi2-xSiyP3-yO12) glass is a good lithium ion conductor. However, the interaction between LATP glass and H+ ions (including its diffusion and surface adsorption) needs to be well understood before the long-term application of LATP glass in an aqueous electrolyte based Li-air batteries where H+ always present. In this work, we investigate the H+ ion diffusion properties in LATP glass and their surface interactions using both experimental and modeling approaches. Our analysis indicates that the apparent H+ related current observed in the initial cyclic voltammetry scan should be attributed to the adsorption of H+ ions on the LATP glass rather than the bulk diffusion of H+ ions in the glass. Furthermore, the density functional theory calculations indicate that the H+ ion diffusion energy barrier (3.21 eV) is much higher than that of Li+ ion (0.79 eV) and Na+ ion (0.79 eV) in NASICON type LiTi2(PO4)3 material. As a result, the H+ ion conductivity in LATP glass is negligible at room temperature. However, significant surface corrosion was found after the LATP glass was soaked in strong alkaline electrolyte for extended time. Therefore, appropriate electrolytes have to be developed to prevent the corrosion of LATP glass beforemore » its practical application for Li-air batteries using aqueous electrolyte.« less

The effects of different cell positive cans and separators on first-cycle Coulombic efficiency and long-term cycling stability of a high-voltage spinel cathode are investigated systematically. Compared to stainless steel (SS) positive cans, aluminum (Al)-clad SS-316 positive cans are much more resistant to oxidation at high voltages; therefore, the initial Coulombic efficiency of the batteries with Al-clad can is improved by more than 13%. Among the five separators studied in this work, the polyethylene (PE) separator exhibits the best electrochemical stability. The cells using LiCr0.05Ni0.45Mn1.5O4 as the cathode, an Al-clad positive can, and a PE separator exhibits a first-cycle Coulombic efficiency of about 90% and a capacity fading of only 0.01% per cycle.

A patterned coreeshell tin (Sn) nanorod anode is fabricated by pulse electrodeposition of Sn onto a selfassembled Tobacco Mosaic Virus (TMV) structured nickel current collector. Pulse electrodeposition onto the virus assembled 3D electrode surfaces produces homogenous Sn coatings with significant void space to accommodate the large volume change associated with Sn lithiation. The TMV enabled 3D Sn anodes shows high capacity retention of 560 mAh g � 1 after 100 cycles with the average capacity fading rate of 0.4% per cycle. The high electronic conductivity of Sn, short diffusion length for Li-ions, and large interface between Sn nano-rods and electrolyte greatly enhance the rate performance of the TMV enabled Sn anodes.

Different classifiers show different sensitivities to translation-variance. The translation-insensitive classifiers are capable of accelerating the detection process by searching over a coarse grid as well as guaranteeing the recall rate.In this paper, we define a concept of Translation-Tolerable Region (TTR) for a classifier. The TTR is such a region that all the detection windows in it have consistent (stable) results output by the classifier. We use the classifier's Maximal Translation-Tolerable Region (MTTR) to measure its sensitivity to the translation-variance. For object detection, we propose an algorithm for training the discriminative classifiers as well as learning the associated MTTRs. The discriminative classifiers are assembled into a cascaded classifier in descending order of their MTTR sizes. To speed up the detection process, we propose a Granularity-Adaptively-Tunable (GAT) search strategy according to the classifiers' MTTRs. Furthermore, we prove that the recall rate is Probably Approximately Admissible (PAA) in the GAT search, which means that the proposed approach can theoretically guarantee the accuracy while accelerating the detection process.Based on the boosting framework with Histograms of Oriented Gradients (HOG) features, we evaluate the proposed approach on the public datasets containing both rigid and non-rigid object classes. The experimental results show that our approach achieves considerable results with a fast speed. Display Omitted Highlights? We define the Maximal Translation-Tolerable Region (MTTR) to measure classifiers' sensitivity to translation-variance. ? We train a cascaded classifier that the MTTRs of the earlier strong classifiers in it are large and vice versa. ? We propose a Granularity-Adaptively-Tunable (GAT) search according to the strong classifiers' MTTRs. ? The recall rate is probably approximately admissible in the GAT search.

Leukemic stem cells (LSCs) play the central role in the relapse and refractory of acute myeloid leukemia (AML) and highlight the critical need for the new therapeutic strategies to directly target the LSC population. However, relatively little is known about the unique molecular mechanisms of drug and natural killer cells (NK)-killing resistance of LSCs because of very small number of LSCs in bone marrow. In this study, we investigated whether established leukemia cell line contains LSCs. We showed that KG1a leukemia cell line contained leukemic stem-like cells, which have been phenotypically restricted within the CD34(+)CD38(-) fractions. CD34(+)CD38(-) cells could generate CD34(+)CD38(+) cells in culture medium and had renewal function. Moreover, CD34(+)CD38(-) cells had self-renewal potential. We found that leukemic stem-like cells from KG1a cells were resistant to chemotherapy and NK-mediated cytotoxicity. NKG2D ligands involve in protecting LSCs from NK-mediated attack. Taken together, our studies provide a novel cell model for leukemic stem cells research. Our data also shed light on mechanism of double resistant to chemotherapy and NK cell immunotherapy, which was helpful for developing novel effective strategies for LSCs.

A patterned 3D Si anode is fabricated by physical vapor deposition of n-type Si on a self-assembled TMV1cys-structured nickel current collector. The combination of the large surface area conferred by the virus-enabled 3D Ni/TMV1cys current collector with the high electric conductivity of n-type Si rods results in excellent cyclic stability and rate capability for the core-shell n-type Si/Ni/TMV1cys anodes. Electrochemical impedance spectroscopy reveals that the high electronic conductivity of n-type Si significantly reduces charge transfer resistance, thus even at high current densities the capacity of the n-type Si is increased to almost 630 mAh/g compared to undoped Si.

The effects of carbonization process and carbon nanofiber/nanotube additives on the cycling stability of silicon–carbon composite anodes were investigated by monitoring the impedance evolution during charge/discharge cycles with electrochemical impedance spectroscopy (EIS). Three types of Si–C anodes were investigated: the first type consisted of Si nanoparticles incorporated into a network of carbon nanofibers (CNFs) and multi-walled carbon nanotubes (MWNTs), with annealed polymer binder. The second type of Si–C anodes was prepared by further heat treatment of the first Si–C anodes to carbonize the polymer binder. The third Si–C anode was as same as the second one except no CNFs and MWNTs being added. Impedance analysis revealed that the carbonization process stabilized the Si–C anode structure and decreased the charge transfer resistance, thus improving the cycling stability. On the other hand, although the MWNTs/CNFs additives could enhance the electronic conductivity of the Si–C anodes, the induced inhomogeneous structure decreased the integrity of the electrode, resulting in a poor long term cycling stability.