Search

An integrated battery system, which integrates solar power and rechargeable battery in the same unit, is recognized as a propective solution for the shortage and inefficiency of power energy. Herein, a hybrid S/N719 dye cathode is proposed in the rechargeable Li-S battery to realize the photo-assisted charging battery system. Specifically, the photo-charge contribution reduces the charge voltage of the sulfur cathode by 0.12 V and accelerates the charge process under light illumination, showing the great potential in saving electric energy and expediting fast-charge capacity of Li-S battery.

It is easy to lead to poor generalization in machine learning tasks using real-world data directly, since such data is usually high-dimensional dimensionality and limited. Through learning the low dimensional representations of high-dimensional data, feature selection can retain useful features for machine learning tasks. Using these useful features effectively trains machine learning models. Hence, it is a challenge for feature selection from high-dimensional data. To address this issue, in this paper, a hybrid approach consisted of an autoencoder and Bayesian methods is proposed for a novel feature selection. Firstly, Bayesian methods are embedded in the proposed autoencoder as a special hidden layer. This of doing is to increase the precision during selecting non-redundant features. Then, the other hidden layers of the autoencoder are used for non-redundant feature selection. Finally, compared with the mainstream approaches for feature selection, the proposed method outperforms them. We find that the way consisted of autoencoders and probabilistic correction methods is more meaningful than that of stacking architectures or adding constraints to autoencoders as regards feature selection. We also demonstrate that stacked autoencoders are more suitable for large-scale feature selection, however, sparse autoencoders are beneficial for a smaller number of feature selection. We indicate that the value of the proposed method provides a theoretical reference to analyze the optimality of feature selection.

In large smelting plants such as steel plant and coking plant the utilization rate of the byproduct gas is generally low, especially in developing countries. Usually, multiple low-parameter boilers are combined with multiple low-parameter steam turbines, which are built in phases, to drive the smelting accessory machineries. In terms of energy utilization efficiency, this model is definitely not economical. Based on the law of thermodynamics, the direction of efficiency improvement was proposed, and the solution of gas utilization system was put forward, according to the Rankine cycle principle. Finally, with a steel plant as the research object, the proposed scheme was applied in the transformation of this plant, and the economic benefit was estimated. The results show that the proposed scheme in this paper can generate more electricity with the same gas consumption.

In the control of air pollutant emissions, the measured emission concentration should be converted into the standard condition. The converted concentration is obtained in order to determine whether the emissions meet requirements. At present, there are several conversion methods available, and the EAR (excess air ratio) conversion method is one of them. Based on the conversion mechanism of air pollutant concentration, the origin of EAR conversion method was deduced and analyzed in this paper. The results show that the EAR method is a simplified method that is only suitable for the fuels with fuel composition factor close to zero, such as coal, biofuel, solid waste, oil, etc. It is not suitable for the fuels with big fuel composition factor, such as blast furnace gas, converter gas and producer gas, etc.

BackgroundPositive symptoms are a useful predictor of aggression in schizophrenia. Although a similar pattern of abnormal brain structures related to both positive symptoms and aggression has been reported, this observation has not yet been confirmed in a single sample.MethodTo study the association between positive symptoms and aggression in schizophrenia on a neurobiological level, a prospective meta-analytic approach was employed to analyze harmonized structural neuroimaging data from 10 research centers worldwide. We analyzed brain MRI scans from 902 individuals with a primary diagnosis of schizophrenia and 952 healthy controls.ResultsThe result identified a widespread cortical thickness reduction in schizophrenia compared to their controls. Two separate meta-regression analyses revealed that a common pattern of reduced cortical gray matter thickness within the left lateral temporal lobe and right midcingulate cortex was significantly associated with both positive symptoms and aggression.ConclusionThese findings suggested that positive symptoms such as formal thought disorder and auditory misperception, combined with cognitive impairments reflecting difficulties in deploying an adaptive control toward perceived threats, could escalate the likelihood of aggression in schizophrenia.

This work presents a miniaturised liquid crystal (LC) phase shifter operating at 6GHz, with a glass-liquid crystal-glass structure for the phase shifter part and slot-coupled structures for the input and output ports. By varying the bias voltage loaded on both sides of the LC material, a phase shift range of 220°can be achieved with a maximum FoM 78.8°/dB.

Weakly-supervised temporal action localization aims to locate intervals of action instances with only video-level action labels for training. However, the localization results generated from video classification networks are often not accurate due to the lack of temporal boundary annotation of actions. Our motivating insight is that the temporal boundary of action should be stably predicted under various temporal transforms. This inspires a self-supervised equivariant transform consistency constraint. We design a set of temporal transform operations, including naive temporal down-sampling to learnable attention-piloted time warping. In our model, a localization network aims to perform well under all transforms, and another policy network is designed to choose a temporal transform at each iteration that adversarially brings localization result inconsistent with the localization network's. Additionally, we devise a self-refine module to enhance the completeness of action intervals harnessing temporal and semantic contexts. Experimental results on THUMOS14 and ActivityNet demonstrate that our model consistently outperforms the state-of-the-art weakly-supervised temporal action localization methods.

Visual backdoor attack is a recently-emerging task which aims to implant trojans in a deep neural model. A trojaned model responds to a trojan-invoking trigger in a fully predictable manner while functioning normally otherwise. As a key motivating fact to this work, most triggers adopted in existing methods, such as a learned patterned block that overlays a benigh image, can be easily noticed by human. In this work, we take image recognition and detection as the demonstration tasks, building trojaned networks that are significantly less human-perceptible and can simultaneously attack multiple targets in an image. The main technical contributions are two-folds: first, under a relaxed attack mode, we formulate trigger embedding as an image steganography-and-steganalysis problem that conceals a secret image in another image in a decipherable and almost invisible way. In specific, a variable number of different triggers can be encoded into a same secret image and fed to an encoder module that does steganography. Secondly, we propose a generic split-and-merge scheme for training a trojaned model. Neurons are split into two sets, trained either for normal image recognition / detection or trojaning the model. To merge them, we novelly propose to hide trojan neurons within the nullspace of the normal ones, such that the two sets do not interfere with each other and the resultant model exhibits similar parameter statistics to a clean model. Comprehensive experiments are conducted on the datasets PASCAL VOC and Microsoft COCO (for detection) and a subset of ImageNet (for recognition). All results clearly demonstrate the effectiveness of our proposed visual trojan method.

An optical see-through two-dimensional (2D)/three-dimensional (3D) compatible display using variable-focus lens and multiplexed holographic optical elements (MHOE) is presented. It mainly consists of a MHOE, a variable-focus lens and a projection display device. The customized MHOE, by using the angular multiplexing technology of volumetric holographic grating, records the scattering wavefront and spherical wavefront array required for 2D/3D compatible display. In particular, we proposed a feasible method to switch the 2D and 3D display modes by using a variable-focus lens in the reconstruction process. The proposed system solves the problem of bulky volume, and makes the MHOE more efficient to use. Based on the requirements of 2D and 3D displays, we calculated the liquid pumping volume of the variable-focus lens under two kinds of diopters.

This paper presents a wide-beam 3D-fractal Hilbert GNSS antenna in the RDSS-S band. The proposed antenna consists of a novel radiator with four two-order 3D-fractal Hilbert curve radiating arms, and the corresponding feeding circuit. The beamwidth, with the gain greater than 0dB of the antenna, is 148° (Phi=0°) and 138° (Phi=90°). The size of the antenna is 30mm×30mm×12mm. In addition, the model of the 3D-fractal Hilbert curve and a prototype of the two-order RDSS-S antenna were manufactured by 3D printing. Simulated and measured results show that the proposed wide-beam 3D-Fractal Hilbert antenna is very suitable for GNSS.

In this paper, we propose a novel model with a hierarchical photo-scene encoder and a reconstructor for the task of album storytelling. The photo-scene encoder contains two sub-encoders, namely the photo and scene encoders, which are stacked together and behave hierarchically to fully exploit the structure information of the photos within an album. Specifically, the photo encoder generates semantic representation for each photo while exploiting temporal relationships among them. The scene encoder, relying on the obtained photo representations, is responsible for detecting the scene changes and generating scene representations. Subsequently, the decoder dynamically and attentively summarizes the encoded photo and scene representations to generate a sequence of album representations, based on which a story consisting of multiple coherent sentences is generated. In order to fully extract the useful semantic information from an album, a reconstructor is employed to reproduce the summarized album representations based on the hidden states of the decoder. The proposed model can be trained in an end-to-end manner, which results in an improved performance over the state-of-the-arts on the public visual storytelling (VIST) dataset. Ablation studies further demonstrate the effectiveness of the proposed hierarchical photo-scene encoder and reconstructor., 8 pages, 4 figures

In this paper, we propose a novel local and global deep matching model to tackle bidirectional image-sentence retrieval. Our proposed matching model can simultaneously exploit the image representation, sentence representation, as well as their complicated matching relationships from both local and global perspectives. For images, two different convolutional neural networks (CNNs) are leveraged to encode the local and global contents, with selective attentions to the image sub-regions and the whole image. For sentences, a CNN based sentence model and Fisher vector are employed to capture the global and local semantic meanings, respectively. Relying on the local and global representations of the image and sentence, the proposed deep matching model learns the complicated image-sentence matching relationships from local and global perspectives by integrating cross-modality correlations with intra-modality similarities. Extensive experimental results demonstrate that the proposed local and global matching model outperforms the state-of-the-art bidirectional retrieval approaches on the Flickr8K, Flickr30K, and MSCOCO datasets. Moreover, the image and sentence representations exploited in local and global levels are demonstrated to play synergic and complementary roles for bidirectional image-sentence retrieval.

Domain adaptation is the situation for supervised learning in which the training data are sampled from the source domain while the test data are sampled from the target domain that follows a different distribution. The key to solving such a problem is to reduce effects of the discrepancy between the training data and test data. Recently, deep learning methods that employ stacked denoising auto-encoders (SDAs) to learn new representations for both domains have been successfully applied in domain adaptation. And, remarkable performance on multi-domain sentiment analysis datasets has been reported, making deep learning a promising approach to domain adaptation problems. In this paper, a deep learning method called Stacked Robust Adaptively Regularized Auto-regressions (SRARAs) is proposed to learn useful representations for domain adaptation problems. Each layer of SRARAs contains two steps: a linear transformation step, which is based on robust adaptively regularized auto-regression, and a non-linear squashing transformation step. The first step aims at reducing the discrepancy between the training data and test data, and the second step is to introduce non-linearity and control the range of the elements in the outputs. The experimental results on text and image datasets demonstrate that the proposed method is very effective.

With the availability of abundant online multi-relational video information, recommender systems that can effectively exploit these sorts of data and suggest creatively interesting items will become increasingly important. Recent research illustrates that tensor models offer effective approaches for complex multi-relational data learning and missing element completion. So far, most tensor-based user clustering models have focused on the accuracy of recommendation. Given the dynamic nature of online media, recommendation in this setting is more challenging as it is difficult to capture the users’ dynamic topic distributions in sparse data settings as well as to identify unseen items as candidates of recommendation. Targeting at constructing a recommender system that can encourage more creativity, a deep Bayesian probabilistic tensor framework for tag and item recommendation is proposed. During the score ranking processes, a metric called Bayesian surprise is incorporated to increase the creativity of the recommended candidates. The new algorithm, called Deep Canonical PARAFAC Factorization (DCPF), is evaluated on both synthetic and large-scale real-world problems. An empirical study for video recommendation demonstrates the superiority of the proposed model and indicates that it can better capture the latent patterns of interactions and generates interesting recommendations based on creative tag combinations.

Background: Few studies have been conducted on psychosomatic health status of pharmacy staff during the COVID-19 pandemic. This study aims to investigate the incidence and influence factors of psychosomatic syndrome of pharmacy staff during the COVID-19 pandemic. Methods: A total of 10721 pharmacy staff received online investigation through a period of 22 days from February 24th to March 16th 2019. The investigation included the self-designed general situation questionnaire and Psychosomatic Symptoms Scale (PSSS), and 9118 participants provided valid questionnaire feedback. ANOVA was used to evaluate significant differences of psychosomatic syndromes in different subgroups. Multiple stepwise linear regression analysis was used to determine the main risk factors of psychosomatic syndrome.Results: During the outbreak of COVID-19, the total incidence of psychosomatic syndrome was 21.7% in the pharmacy staff. The most common psychosomatic symptoms were sleep problems (dyscoimesis) and mood problems (irritability). Age was the most important risk factor of the observed psychosomatic syndromes and somatic symptoms, and education was identified affecting mostly psychological symptoms.Conclusion: During the period of COVID-19, the psychosomatic problems of pharmacy staff were prominent. Age and educational background should be taken into account of potential intervention strategy. The relief of mood and sleep will aid the treatment effort.

Immune checkpoint blockade (ICB) has a limited effect on colorectal cancer, underlining the requirement of co‐targeting the complementary mechanisms. Here, we identified prostaglandin E2 (PGE2) receptor 4 (EP4) as the master regulator of immunosuppressive myeloid cells (IMCs), which are the major driver of resistance to ICB therapy. PGE2‐bound EP4 promotes the differentiation of immunosuppressive M2 macrophages and myeloid‐derived suppressor cells (MDSCs) and reduces the expansion of immunostimulated M1 macrophages. To explore the immunotherapeutic role of EP4 signaling, we developed a novel and selective EP4 antagonist TP‐16. TP‐16 effectively blocked the function of IMCs and enhanced cytotoxic T‐cell‐mediated tumor elimination in vivo. Cell co‐culture experiments revealed that TP‐16 promoted T‐cell proliferation, which was impaired by tumor‐derived CD11b+ myeloid cells. Notably, TP‐16 and anti‐PD‐1 combination therapy significantly impeded tumor progression and prolonged mice survival. We further demonstrated that TP‐16 increased responsiveness to anti‐PD‐1 therapy in an IMC‐related spontaneous colorectal cancer mouse model. In summary, this study demonstrates that inhibition of EP4‐expressing IMCs may offer a potential strategy for enhancing the efficacy of immunotherapy for colorectal cancer., Immunosuppressive myeloid cells (IMCs) are a prominent driver of immunotherapy resistance in colorectal cancer. This study identifies EP4 as a master regulator of IMCs and highlights blockade of EP4 as a novel therapeutic strategy for enhancing immunotherapy in colorectal cancer.

Large animal model of optic nerve crush (ONC) plays an important role in translating novel therapeutic strategies developed in rodent model to clinical application. Due to the poor accessibility of the optic nerve (ON) in humans and large animals, lateral orbitotomy is needed to expose the retrobulbar ON. This study was to explore the effects of ONC and ON exposure with lateral orbitotomy (sham surgery) on the outer retinal function and structure in goats by using standard flash electroretinogram (FERG) and spectral-domain optical coherence tomography (SD-OCT). We found that ONC led to a transient reduction in FERG amplitudes at 1 week post injury (wpi), which recovered gradually over 2 months afterwards. Sham surgery alone also caused a similar pattern of amplitude reduction in FERG, although not as significantly as ONC did. Transient outer retinal thickening following ONC occurred at 4 wpi (when progressive thinning of the ganglion cell complex began), peaked at 8 wpi, then recovered gradually at 12 wpi. In contrast, outer retinal thickness remained unchanged statistically 3 months after sham surgery. Fundus fluorescein angiography showed that neither ONC nor ON exposure with lateral orbitotomy significantly caused any significant delay or absence of central retinal vascular filling. In summary, ONC with lateral orbitotomy affects outer retinal function and structure transiently.

Currently, various treatment options are available for lower urinary tract symptoms (LUTS) secondary to benign prostatic hyperplasia (BPH), mainly including medical therapy, open prostatectomy (OP), transurethral resection of prostate (TURP), and minimally invasive treatment, such as holmium laser enucleation (HoLEP), greenlight laser prostatectomy, photovaporization of the prostate, etc. [1–4]. Medical therapy, including the use of alpha-1-Blocker and 5-alpha reductase inhibitors, has been considered first-line treatment [5]. Surgical treatments are used for treatment in patients who are failed in medical treatment [6]. European Association of Urology suggests that TURP is considered as the standard therapy for patients with prostate volume (PV) ranges from 30 to 80 cm3, OP or HoLEP are considered as the gold standard when PV larger than 80 cm3 [2, 3, 7]. However, each treatment method has its advantages and disadvantages. The medication may affect blood pressure and sexual function. Surgical treatment is associated with prolonged hospitalization, postoperative pain, retrograde ejaculation, sexual dysfunction, and hemorrhage [3, 8–10].

Attention-deficit/hyperactivity disorder (ADHD) is a childhood-onset neuropsychiatric disorder, and its existence in adulthood is well established. Beyond symptoms of inattention and hyperactivity/impulsivity, patients commonly present with impairments in cognition. How neuronal underpinnings of symptoms and cognitive deficits differ across adolescence and adulthood is not clear. In this cross sectional study, we investigated gray matter of two cohorts, 486 adults and 508 adolescents, each including participants with ADHD and healthy controls. Independent component analysis was applied to the gray matter of each cohort, separately, to extract cohort specific networks. Then, we identified gray matter networks associated with symptoms, working memory and/or diagnosis in each cohort, and projected them onto the other cohort for comparison. Two components in the inferior, middle/superior frontal regions identified in adults and one component in the insula and inferior frontal region identified in adolescents were significantly associated with working memory deficits in both cohorts. One component in bilateral cerebellar tonsil and culmen identified in adults and one component in left cerebellar region identified in adolescents were significantly associated with inattentive symptoms in both cohorts. All these components presented significant or nominal level of gray matter reduction for ADHD patients in adolescents, but only one showed nominal reduction for patients in adults. Our findings suggest gray matter reduction may not be a sensitive marker for persist ADHD. However, the patterns of certain brain regions are associated with deficits in working memory or attention persistently from childhood into adulthood, which might help understand the mechanism of disease persistence.

Background Several studies have demonstrated autophagy was involved in the process of esophageal adenocarcinoma (EAC). The aim of this study was to explore autophagy-related genes (ARGs) correlated with overall survival (OS) in EAC patients. Methods Expressions of ARGs in EAC and normal samples were downloaded from TCGA database. GO and KEGG enrichment analyses were used to investigate the ARGs bioinformatics functions. Univariate and multivariate cox regressions were performed to identify prognostic ARGs and the independent risk factors. ROC curve was established to evaluate the feasibility to predict the prognosis. Finally, the correlations between ARGs and clinical features were further explored. In addition, significantly different ARGs were verified in EAC specimens and normal esophageal mucosal tissues. Results Thirty significantly different ARGs were selected from EAC and normal tissues. Functional enrichments showed these ARGs were mainly related apoptosis. Multivariate cox regression analyses demonstrated eight ARGs were significantly associated with OS. Among these eight genes, BECN1 (HR = 0.321, P = 0.046), DAPK1 (HR = 0.636, P = 0.025) and CAPN1 (HR = 0.395, P = 0.004) played protective roles in survival. Gender (HR = 0.225, P = 0.032), stage (HR = 5.841, P = 0.008) and risk score (HR = 1.131, P < 0.001) were independent prognostic risk factors. ROC curves showed better efficacy to predict survival using the risk score. Additionally, we found BECN1, DAPK1, VAMP7 and SIRT1 genes were correlated significantly with survival status, gender, primary tumor and tumor stage (all P < 0.05). The experimental results confirmed the BIRC5 was overexpressed and the ITPR1, PRKN were downregulated in the EAC tissues compared with the normal esophageal mucosal tissues (all P < 0.05). Conclusion Our findings suggested that autophagy was involved in the process of EAC. Several ARGs probably could serve as diagnostic and prognostic biomarkers and may help facilitate therapeutic targets in EAC patients.

Bipolar disorder (BD) is a severe manic-depressive illness. Patients with BD have been shown to have gray matter (GM) deficits in prefrontal, frontal, parietal, and temporal regions; however, the relationship between structural effects and clinical profiles has proved elusive when considered on a region by region or voxel by voxel basis. In this study, we applied parallel independent component analysis (pICA) to structural neuroimaging measures and the positive and negative syndrome scale (PANSS) in 110 patients (mean age 34.9 ± 11.65) with bipolar disorder, to examine networks of brain regions that relate to symptom profiles. The pICA revealed two distinct symptom profiles and associated GM concentration alteration circuits. The first PANSS pICA profile mainly involved anxiety, depression and guilty feelings, reflecting mood symptoms. Reduced GM concentration in right temporal regions predicted worse mood symptoms in this profile. The second PANSS pICA profile generally covered blunted affect, emotional withdrawal, passive/apathetic social withdrawal, depression and active social avoidance, exhibiting a withdrawal or apathy dominating component. Lower GM concentration in bilateral parietal and frontal regions showed worse symptom severity in this profile. In summary, a pICA decomposition suggested BD patients showed distinct mood and apathy profiles differing from the original PANSS subscales, relating to distinct brain structural networks., Highlights • Structural relationships with symptoms in bipolar disorder are complex. • A parallel ICA analysis of PANSS questions and structural images finds two correlated profiles. • The first pair links mood symptoms with right temporal regions. • The second pair highlights social withdrawal and apathy symptoms linked to bilateral frontal and parietal regions.

Objective: We propose and develop a novel biclustering (N-BiC) approach for performing N-way biclustering of neuroimaging data. Our approach is applicable to an arbitrary number of features from both imaging and behavioral data (e.g., symptoms). We applied it to structural MRI data from patients with schizophrenia. Methods: It uses a source-based morphometry approach [i.e., independent component analysis of gray matter segmentation maps] to decompose the data into a set of spatial maps, each of which includes regions that covary among individuals. Then, the loading parameters for components of interest are entered to an exhaustive search, which incorporates a modified depth-first search technique to carry out the biclustering, with the goal of obtaining submatrices where the selected rows (individuals) show homogeneity in their expressions of selected columns (components) and vice versa. Results: Findings demonstrate that multiple biclusters have an evident association with distinct brain networks for the different types of symptoms in schizophrenia. The study identifies two components: inferior temporal gyrus (16) and brainstem (7), which are related to positive (distortion/excess of normal function) and negative (diminution/loss of normal function) symptoms in schizophrenia, respectively. Conclusion: N-BiC is a data-driven method of biclustering MRI data that can exhaustively explore relationships/substructures from a dataset without any prior information with a higher degree of robustness than earlier biclustering applications. Significance: The use of such approaches is important to investigate the underlying biological substrates of mental illness by grouping patients into homogeneous subjects, as the schizophrenia diagnosis is known to be relatively nonspecific and heterogeneous.

Highlights • Brain mechanisms behind familial effects and ADHD symptoms are examined in children. • Genetic liability and symptoms are distinctly reflected by structural brain areas. • The cerebellum is negative correlated with inattention across the entire sample. • The insula in siblings and cases showed reduced loadings compared to controls., Gray matter disruptions have been found consistently in Attention-deficit/Hyperactivity Disorder (ADHD). The organization of these alterations into brain structural networks remains largely unexplored. We investigated 508 participants (281 males) with ADHD (N = 210), their unaffected siblings (N = 108), individuals with subthreshold ADHD (N = 49), and unrelated healthy controls (N = 141) with an age range from 7 to 18 years old from 336 families in the Dutch NeuroIMAGE project. Source based morphometry was used to examine structural brain network alterations and their association with symptoms and cognitive performance. Two networks showed significant reductions in individuals with ADHD compared to unrelated healthy controls after False Discovery Rate correction. Component A, mainly located in bilateral Crus I, showed a ADHD/typically developing difference with subthreshold cases being intermediate between ADHD and typically developing controls. The unaffected siblings were similar to controls. After correcting for IQ and medication status, component A showed a negative correlation with inattention symptoms across the entire sample. Component B included a maximum cluster in the bilateral insula, where unaffected siblings, similar to individuals with ADHD, showed significantly reduced loadings compared to controls; but no relationship with individual symptoms or cognitive measures was found for component B. This multivariate approach suggests that areas reflecting genetic liability within ADHD are partly separate from those areas modulating symptom severity.

Automatically generating sentences to describe events and temporally localizing sentences in a video are two important tasks that bridge language and videos. Recent techniques leverage the multimodal nature of videos by using off-the-shelf features to represent videos, but interactions between modalities are rarely explored. Inspired by the fact that there exist cross-modal interactions in the human brain, we propose a novel method for learning pairwise modality interactions in order to better exploit complementary information for each pair of modalities in videos and thus improve performances on both tasks. We model modality interaction in both the sequence and channel levels in a pairwise fashion, and the pairwise interaction also provides some explainability for the predictions of target tasks. We demonstrate the effectiveness of our method and validate specific design choices through extensive ablation studies. Our method turns out to achieve state-of-the-art performances on four standard benchmark datasets: MSVD and MSR-VTT (event captioning task), and Charades-STA and ActivityNet Captions (temporal sentence localization task).

Many real-world applications propose the request for sharing knowledge among different tasks or datasets. Transfer learning has been proposed to solve this kind of problems and it has been successfully applied in supervised learning and semi-supervised learning settings. However, its adoption in clustering, one of the most classical research problems in machine learning and data mining, is still scarce. Spectral clustering, as a major clustering algorithm with wide applications and better performance than k-means typically, has not been well incorporated with knowledge transfer. In this paper, we first consider the problem of learning from only one auxiliary unlabeled dataset for spectral clustering and propose a novel algorithm called transfer spectral clustering (TSC). Then, it is extended to the settings with multiple auxiliary tasks. TSC assumes the feature embeddings being shared with the auxiliary tasks and utilizes co-clustering to extract useful information from the auxiliary datasets to improve the clustering performance. TSC involves not only the data manifold information of individual task but also the feature manifold information shared between related tasks. An in-depth explanation of our algorithm together with a convergence analysis are provided. As demonstrated by the extensive experiments, TSC can effectively improve the clustering performance by using auxiliary unlabeled data when compared with other state-of-the-art clustering algorithms.

While gray matter (GM) anomalies have been reported for attention-deficit/hyperactivity disorder (ADHD), investigating their associations with cognitive deficits and individual symptom domains can help pinpoint the neural underpinnings critical for the pathology of ADHD, particularly the persist form of ADHD. In this work, we performed both independent component analysis and voxel-based morphometry analysis on whole brain GM of 486 adults including 214 patients, 96 unaffected siblings, and 176 healthy controls, in relation to cognition and symptoms. Independent component analysis revealed that higher GM volume in inferior semilunar lobule, inferior frontal gyri, and superior and middle frontal gyri was associated with better working memory performance, and lower GM volume in cerebellar tonsil and culmen was associated with more severe inattention symptoms. Consistently, voxel-based morphometry analysis showed that higher GM volume in multiple regions of frontal lobe, cerebellum and temporal lobe was related to better working memory performance. Focusing on the networks derived from ICA, our results integrated prefrontal regions and cerebellar regions through associations with working memory and inattention symptoms, lending support for the theory of ‘cool’-cognition dysfunction being mediated by inferior fronto-striato-cerebellar networks in ADHD. Siblings showed intermediate cognitive impairments between patients and controls but presented GM anomalies in unique focal regions, suggesting they are a separate group potentially affected by the shared genetic and environmental risks with ADHD patients., Highlights • Frontal and cerebellar regions associated with adult ADHD working memory deficits. • Gray matter reduction in cerebellum related to inattention symptoms in adult ADHD. • Siblings presented intermediate cognitive impairments. • Adult ADHD showed gray matter reduction in middle frontal gyrus.

The task of temporally grounding language queries in videos is to temporally localize the best matched video segment corresponding to a given language (sentence). It requires certain models to simultaneously perform visual and linguistic understandings. Previous work predominantly ignores the precision of segment localization. Sliding window based methods use predefined search window sizes, which suffer from redundant computation, while existing anchor-based approaches fail to yield precise localization. We address this issue by proposing an end-to-end boundary-aware model, which uses a lightweight branch to predict semantic boundaries corresponding to the given linguistic information. To better detect semantic boundaries, we propose to aggregate contextual information by explicitly modeling the relationship between the current element and its neighbors. The most confident segments are subsequently selected based on both anchor and boundary predictions at the testing stage. The proposed model, dubbed Contextual Boundary-aware Prediction (CBP), outperforms its competitors with a clear margin on three public datasets. All codes are available on https://github.com/JaywongWang/CBP ., Accepted to AAAI 2020

White matter magnetic resonance hyperintensities of presumed vascular origin, which could be widely observed in elderly people, and has significant importance in multiple neurological studies. Quantitative measurement usually relies heavily on manual or semi-automatic delineation and intuitive localization, which is time-consuming and observer-dependent. Current automatic quantification methods focus mainly on the segmentation, but the spatial distribution of lesions plays a vital role in clinical diagnosis. In this study, we implemented four segmentation algorithms and compared the performances quantitatively and qualitatively on two open-access datasets. The location-specific analysis was conducted sequentially on 213 clinical patients with cerebral ischemia and lacune. The experimental results suggest that our deep-learning-based model has the potential to be integrated into the clinical workflow.

In this paper, we propose a reflective augmented reality (AR) display system based on integral imaging (II) using a mirror-based pinhole array (MBPA). The MBPA, obtained by punching pinholes on a mirror, functions as a three-dimensional (3D) imaging device, as well as an image combiner. The pinhole array of MBPA can realize a pinhole array-based II display, while the mirror of MBPA can image the real objects, so as to combine the images of the real objects with the reconstructed 3D images. The structure of the proposed reflective AR display is very simple, and only a projection system or a two-dimensional display screen is needed to combine with the MBPA. In our experiment, a 25cm &times, 14cm sized AR display was built up, a combination of a 3D virtual image and a real 3D object was presented by the proposed AR 3D display. The proposed device could realize an AR display of large size due to its compact form factor and low weight.

Current diagnoses of schizophrenia and related psychiatric disorders are classified by phenomenological principles and clinical descriptions while ruling out other symptoms and conditions. Specific biomarkers are needed to assist the current diagnostic system. However, complicated gene and environmental factor interactions play an important role in the disease onset and induce great heterogeneity. This unclear etiology and heterogeneity raise the difficulty to induce the common biomarkers to distinguish schizophrenia from the healthy population. Simultaneously, the vast overlap in common symptoms, genetic variations and brain alterations in schizophrenia and related psychiatric disorders indicate the extension of biomarkers only achieved in schizophrenia. This complex situation compels the development of biomarkers based on research findings not only in schizophrenia but also with related disorders that includes behavior as well as imaging genetics. Imaging genetics is a unique methodology to assess the impact of genetic factors on both brain structure and function. More importantly, imaging genetics builds a bridge to understand the behavioral and clinical implications of genetics and neuroimaging. By characterizing and quantifying the brain affected in psychiatric disorders, imaging genetics is contributing to present potential biomarkers for schizophrenia and related disorders. To date, imaging genetics research includes candidate gene analysis, genome-wide association studies, polygenetic risk score analysis, and large scale collaborative studies that have made contributions to the understanding of schizophrenia with the potential to serve as biomarkers. Though limitations such as failing to capture enough genetic information and methodological arguments regarding imaging genetics itself may delay its progress, imaging genetics remains a promising as more aggregative, clustering methods and imaging genetics-compatible clinical assessments are employed in future studies.

Early diagnosis of prostate cancer is very crucial to reduce the mortality rate. Multi-parametric magnetic resonance imaging (MRI) can provide detailed visualization of prostate tissues and lesions. Their malignancy can be diagnosed before any necessary invasive approaches, such as needle biopsy, at the risk of damage to or inflammation of the periprostatic nerves, prostate and bladder neck. However, the prostate tissue malignancy on magnetic resonance (MR) images can also be difficult to determine, with often inconclusive results among the clinicians. With the progress in artificial intelligence (AI), research on MR image-based lesion classification with AI tools are being explored increasingly. So far, existing classification approaches heavily rely on manually labelling of lesion areas, which is a labor-intensive and time-consuming process. In this paper, we present a novel two-stage method for fully-automated prostate lesion detection and classification, using input sequences of T2-weighted images, apparent diffusion coefficient (ADC) maps and high b-value diffusion-weighted images. In the first stage, a Mask R-CNN model is trained to automatically segment prostate structures. In the second stage, a weakly supervised deep neural network is developed to detect and classify lesions in a single run. To validate the accuracy of our system, we tested our method on two datasets, one from the PROSTATEx Challenge and the other from our local cohort. Our method can achieve average area-under-the-curve (AUC) of 0.912 and 0.882 on the two datasets respectively. The proposed approach present a promising tool for radiologists in their clinical practices.

The excess air ratio indicates the excess of supporting air for fuel combustion, and is an important control parameter in operation adjustment for combustion equipment. In the present methods for calculating excess air ratio, the conventional method is simple but imprecise, and the improved method is accurate but complicated. Based on the definition and solution model of excess air volume, a new method for calculating excess air ratio was derived, and the concept of fuel component factor was proposed. Finally, the new method was compared with the current methods through fuel composition and combustion data. The results show that the conventional method is only suitable for the fuel whose fuel component factor is close to zero. The new method is obtained by fuel composition factor and flue gas composition, which can not only guarantee the accuracy of the solution, but also simplify the solution process, and it is a general calculation model that can be applied to any fuel.

In this paper, we propose to guide the video caption generation with Part-of-Speech (POS) information, based on a gated fusion of multiple representations of input videos. We construct a novel gated fusion network, with one particularly designed cross-gating (CG) block, to effectively encode and fuse different types of representations, e.g., the motion and content features of an input video. One POS sequence generator relies on this fused representation to predict the global syntactic structure, which is thereafter leveraged to guide the video captioning generation and control the syntax of the generated sentence. Specifically, a gating strategy is proposed to dynamically and adaptively incorporate the global syntactic POS information into the decoder for generating each word. Experimental results on two benchmark datasets, namely MSR-VTT and MSVD, demonstrate that the proposed model can well exploit complementary information from multiple representations, resulting in improved performances. Moreover, the generated global POS information can well capture the global syntactic structure of the sentence, and thus be exploited to control the syntactic structure of the description. Such POS information not only boosts the video captioning performance but also improves the diversity of the generated captions. Our code is at: https://github.com/vsislab/Controllable_XGating., Comment: Accepted by ICCV 2019

Schizophrenia is a complex disorder that affects perception, cognition, and emotion causing symptoms such as delusions, hallucinations, and suspiciousness. Schizophrenia is also associated with structural cortical abnormalities including lower gray matter (GM) concentration, GM volume, and cortical thickness relative to healthy control individuals. However, the association between GM measures and symptom dimensions in schizophrenia is still not well understood. Here, we applied parallel independent component analysis (pICA), a higher-order statistical approach that identifies covarying patterns within two (or more) data modalities simultaneously, to link covarying brain networks of GM concentration with covarying linear combinations of the positive and negative syndrome scale (PANSS). In a large sample of patients with schizophrenia (n = 337) the association between these two data modalities was investigated. The pICA revealed a distinct PANSS profile characterized by increased delusional symptoms, suspiciousness, hallucinations, and anxiety, that was associated with a pattern of lower GM concentration in inferior temporal gyri and fusiform gyri and higher GM concentration in the sensorimotor cortex. GM alterations replicate previous findings; additionally, applying a multivariate technique, we were able to map a very specific symptom profile onto these GM alterations extending our understanding of cortical abnormalities associated with schizophrenia. Techniques like parallel ICA can reveal linked patterns of alterations across different data modalities that can help to identify biologically-informed phenotypes which might help to improve future treatment targets.

Background Health care professionals are required to maintain accurate health records of patients. Furthermore, these records should be shared across different health care organizations for professionals to have a complete review of medical history and avoid missing important information. Nowadays, health care providers use electronic health records (EHRs) as a key to the implementation of these goals and delivery of quality care. However, there are technical and legal hurdles that prevent the adoption of these systems, such as concerns about performance and privacy issues. Objective This study aimed to build and evaluate an experimental blockchain for EHRs, named HealthChain, which overcomes the disadvantages of traditional EHR systems. Methods HealthChain is built based on consortium blockchain technology. Specifically, three organizations, namely hospitals, insurance providers, and governmental agencies, form a consortium that operates under a governance model, which enforces the business logic agreed by all participants. Every peer node hosts an instance of the distributed ledger consisting of EHRs and an instance of chaincode regulating the permissions of participants. Designated orderers establish consensus on the order of EHRs and then disseminate blocks to peers. Results HealthChain achieves functional and nonfunctional requirements. It can store EHRs in a distributed ledger and share them among different participants. Moreover, it demonstrates superior features, such as privacy preservation, security, and high throughput. These are the main reasons why HealthChain is proposed. Conclusions Consortium blockchain technology can help to build new EHR systems and solve the problems that prevent the adoption of traditional systems.

Large animal models of optic nerve injury are essential for translating novel findings into effective therapies due to their similarity to humans in many respects. However, most current tests evaluating the integrity of retinal ganglion cells (RGCs) and optic nerve (ON) are based on rodent animal models. We aimed to evaluate and optimize the in vivo methods to assess RGCs and ON's function and structure in large animals in terms of reproducibility, simplicity and sensitivity. Both goats and rhesus macaques were employed in this study. By using goats, we found anesthesia with isoflurane or xylazine resulted in different effects on reproducibility of flash visual evoked potential (FVEP) and pattern electroretinogram (PERG). FVEP with the large-Ganzfeld stimulator was significantly more stable than that with mini-Ganzfeld stimulator. PERG with simultaneous binocular stimulation, with superior simplicity over separate monocular stimulation, was appliable in goats due to undetectable interocular crosstalk of PERG signals. After ON crush in goats, some FVEP components, PERG, OCT and PLR demonstrated significant changes, in line with the histological study. By using rhesus macaque, we found the implicit time of PVEP, FVEP and PERG were significantly more reproducible than amplitudes, and OCT and PLR demonstrated small intersession variation. In summary, we established an optimized system to evaluate integrity of RGCs and ON in large animals in vivo, facilitating usage of large animal models of optic nerve diseases.

Late-onset depression (LOD) is regarded as a risk factor or a prodrome of Alzheimer’s disease (AD). Moreover, LOD patients with cognitive deficits have the higher risk of subsequent AD. Thus, it is necessary to understand the neural underpinnings of cognitive deficits and its pathological implications in LOD. Consistent findings show that the default mode network (DMN) is an important and potentially useful brain network for the cognitive deficits in LOD patients. In recent years, genetics has been actively researched as a possible risk factor in the pathogenesis of LOD. So, in this review, we discuss the current research progress on the cognitive deficits and DMN in LOD through a combined view of brain network and genetics. We find that different structural and functional impairments of the DMN might be involved in the etiological mechanisms of different cognitive impairments in LOD patients.

To explore the mechanism of transcranial direct current stimulation (tDCS) on the improved performance of professional rowing athletes. Twelve male professional rowing athletes were randomly divided into two groups (low-stimulation group, 1 mA, n = 6; high-stimulation group, 2 mA, n = 6), and they accepted tDCS for two consecutive weeks while undergoing regular training (20 min each time, five times a week, totally ten times). The assessments of depression, anxiety, executive function, fatigue perception, lactate threshold power (LTP) and isokinetic muscle strength as well as the collection of functional magnetic resonance imaging (fMRI) data were performed at baseline and at follow-up (the end of the fourth week). The voxel-mirrored homotopic connectivity (VMHC) value was calculated in the whole brain. After stimulation, there were significant increases in executive function and athletic performance. Analysis of variance (ANOVA) analysis indicated time factor, stimulation intensity factor had a main effect on LTP and 60RK, respectively. There was no significant difference of VMHC value between the high- and low-stimulation groups at baseline. Comparing with low-stimulation group, significant increased VMHC values of the bilateral middle temporal gyrus (MTG), precentral gyrus and superior frontal gyrus (SFG) were found in high-stimulation group at follow-up. Correlation analyses showed that in high-stimulation group, the VMHC values of bilateral MTG and SFG were both positively correlated with the measures of athletic performance. tDCS may contribute to the improvement of athletic performance in professional rowing athletes, and the increased interhemispheric coordination may be involved in the mechanism of the improved athletic performance.

Recently, caption generation with an encoder-decoder framework has been extensively studied and applied in different domains, such as image captioning, code captioning, and so on. In this paper, we propose a novel architecture, namely Auto-Reconstructor Network (ARNet), which, coupling with the conventional encoder-decoder framework, works in an end-to-end fashion to generate captions. ARNet aims at reconstructing the previous hidden state with the present one, besides behaving as the input-dependent transition operator. Therefore, ARNet encourages the current hidden state to embed more information from the previous one, which can help regularize the transition dynamics of recurrent neural networks (RNNs). Extensive experimental results show that our proposed ARNet boosts the performance over the existing encoder-decoder models on both image captioning and source code captioning tasks. Additionally, ARNet remarkably reduces the discrepancy between training and inference processes for caption generation. Furthermore, the performance on permuted sequential MNIST demonstrates that ARNet can effectively regularize RNN, especially on modeling long-term dependencies. Our code is available at: https://github.com/chenxinpeng/ARNet, Accepted by CVPR 2018

Background: Asthma is a chronic disease appeared to be associated with depression. But the underpinnings of depression in asthma remain unknown. In order to understand the neural mechanisms of depression in asthma, we used cerebral blood flow (CBF) to probe the difference between depressed asthmatic (DA) and non-depressed asthmatic (NDA) patients. Methods: Eighteen DA patients, 24 NDA patients and 57 healthy controls (HC) received pulsed arterial spin labeling (pASL) scan for measuring CBF, resting-state functional magnetic resonance imaging (rs-fMRI) scan, severity of depression and asthma control assessment, respectively. Results: Compared to NDA, DA patients showed increased regional CBF (rCBF) in the right cerebellum posterior lobe. Compared to HC, DA, and NDA patients all showed significantly decreased rCBF in the right cerebellum posterior lobe. Conclusions: We showed the first evidence of altered rCBF in the right cerebellum posterior lobe in asthma using pASL, which appeared to be involved in the neuropathology in asthma. Clinical Trial Registration: An investigation of therapeutic mechanism in asthmatic patients: based on the results of Group Cognitive Behavioral Therapy (Registration number: ChiCTR-COC-15007442) (http://www.chictr.org.cn/usercenter.aspx).

Dense video captioning is a newly emerging task that aims at both localizing and describing all events in a video. We identify and tackle two challenges on this task, namely, (1) how to utilize both past and future contexts for accurate event proposal predictions, and (2) how to construct informative input to the decoder for generating natural event descriptions. First, previous works predominantly generate temporal event proposals in the forward direction, which neglects future video context. We propose a bidirectional proposal method that effectively exploits both past and future contexts to make proposal predictions. Second, different events ending at (nearly) the same time are indistinguishable in the previous works, resulting in the same captions. We solve this problem by representing each event with an attentive fusion of hidden states from the proposal module and video contents (e.g., C3D features). We further propose a novel context gating mechanism to balance the contributions from the current event and its surrounding contexts dynamically. We empirically show that our attentively fused event representation is superior to the proposal hidden states or video contents alone. By coupling proposal and captioning modules into one unified framework, our model outperforms the state-of-the-arts on the ActivityNet Captions dataset with a relative gain of over 100% (Meteor score increases from 4.82 to 9.65)., CVPR2018 spotlight paper

Recently, much advance has been made in image captioning, and an encoder-decoder framework has been adopted by all the state-of-the-art models. Under this framework, an input image is encoded by a convolutional neural network (CNN) and then translated into natural language with a recurrent neural network (RNN). The existing models counting on this framework employ only one kind of CNNs, e.g., ResNet or Inception-X, which describes the image contents from only one specific view point. Thus, the semantic meaning of the input image cannot be comprehensively understood, which restricts improving the performance. In this paper, to exploit the complementary information from multiple encoders, we propose a novel recurrent fusion network (RFNet) for the image captioning task. The fusion process in our model can exploit the interactions among the outputs of the image encoders and generate new compact and informative representations for the decoder. Experiments on the MSCOCO dataset demonstrate the effectiveness of our proposed RFNet, which sets a new state-of-the-art for image captioning.

VGF, a non-acronymic neuropeptide, is important in the pathogenesis of major depressive disorder (MDD) and in the functioning and efficacy of some antidepressant drugs. In this study we assessed whether serum VGF levels change in MDD patients and if antidepressant treatments can restore these changes.We measured serum VGF concentrations using sandwich ELISA in drug-free MDD patients before treatment began (n = 26) and at 8 weeks after antidepressant treatment (n = 26) with escitalopram and duloxetine, two common antidepressants. The severity of depression was assessed with the 17-item Hamilton Depression Rating Scale (HDRS).VGF serum levels were significantly lower in MDD patients compared to controls (P = .002), even after controlling for the effects of age and education (P = .037), and they were reversed by 8 weeks of drug treatment (P.0001). Both escitalopram and duloxetine restored the decreased serum VGF levels (P .05). We observed no correlation between VGF levels and HDRS scores in pre-treatment MDD patients (P = .879).The results suggest that VGF may be implicated in the pathophysiology of MDD and in the mechanisms underlying the action of antidepressants, and serum VGF may be regarded as a trait parameter for MDD.