liangliang ren - Academia.edu (original) (raw)

Papers by liangliang ren

Cornell University - arXiv, Sep 1, 2022

The point cloud based 3D single object tracking (3DSOT) has drawn increasing attention. Lots of b... more The point cloud based 3D single object tracking (3DSOT) has drawn increasing attention. Lots of breakthroughs have been made, but we also reveal two severe issues. By an extensive analysis, we find the prediction manner of current approaches is non-robust, i.e., exposing a misalignment gap between prediction score and actually localization accuracy. Another issue is the sparse point returns will damage the feature matching procedure of the SOT task. Based on these insights, we introduce two novel modules, i.e., Adaptive Refine Prediction (ARP) and Target Knowledge Transfer (TKT), to tackle them, respectively. To this end, we first design a strong pipeline to extract discriminative features and conduct the matching procedure with the attention mechanism. Then, ARP module is proposed to tackle the misalignment issue by aggregating all predicted candidates with valuable clues. Finally, TKT module is designed to effectively overcome incomplete point cloud due to sparse and occlusion issues. We call our overall framework PCET. By conducting extensive experiments on the KITTI and Waymo Open Dataset, our model achieves state-ofthe-art performance while maintaining a lower computational consumption.

Frontiers in Oncology

The majority of occult liver metastases cannot be detected by computed tomography (CT), magnetic ... more The majority of occult liver metastases cannot be detected by computed tomography (CT), magnetic resonance imaging (MRI) or other traditionally morphological imaging approaches since the lesions are too small or they have not yet formed cancer nodules. Gankyrin is a small molecular protein composed of seven ankyrin domains. In this study, the expression of Gankyrin in colorectal cancer (CRC) patients with liver metastases was investigated to determine its prognosis value. Gankyrin expression in CRC patients was initially analyzed using data from The Cancer Genome Atlas (TCGA) database and bioinformatics tools. RT-qPCR, western blotting, immunohistochemistry (IHC) and transwell migration and invasion assays were then performed to verify the expression and function of Gankyrin in CRC cell line, CRC tissues and matched non-tumor tissues of clinical patients. General clinicopathological information including TNM stage as well as preoperative and postoperative imaging results were collec...

PeerJ

Purpose We aimed to establish a cholesterogenic gene signature to predict the prognosis of young ... more Purpose We aimed to establish a cholesterogenic gene signature to predict the prognosis of young breast cancer (BC) patients and then verified it using cell line experiments. Methods In the bioinformatic section, transcriptional data and corresponding clinical data of young BC patients (age ≤ 45 years) were downloaded from The Cancer Genome Atlas (TCGA) database for training set. Differentially expressed genes (DEGs) were compared between tumour tissue (n = 183) and normal tissue (n = 30). By using univariate Cox regression and multi COX regression, a five-cholesterogenic-gene signature was established to predict prognosis. Subgroup analysis and external validations of GSE131769 from the Gene Expression Omnibus (GEO) were performed to verify the signature. Subsequently, in experiment part, cell experiments were performed to further verify the biological roles of the five cholesterogenic genes in BC. Results In the bioinformatic section, a total of 97 upregulated genes and 124 downre...

International Journal of Biological Sciences

Bladder cancer is one of the most common and deadly cancer worldwide. Current chemotherapy has sh... more Bladder cancer is one of the most common and deadly cancer worldwide. Current chemotherapy has shown limited efficacy in improving outcomes for patients. Nitroxoline, an old and widely used oral antibiotic, which was known to treat for urinary tract infection for decades. Recent studies suggested that nitroxoline suppressed the tumor progression and metastasis, especially in bladder cancer. However, the underlying mechanism for anti-tumor activity of nitroxoline remains unclear. Methods: CircRNA microarray was used to explore the nitroxoline-mediated circRNA expression profile of bladder cancer lines. Transwell and wound-healing assay were applied to evaluate the capacity of metastasis. ChIP assay was chosen to prove the binding of promotor and transcription factor. RNA-pulldown assay was performed to explore the sponge of circRNA and microRNA. Results: We first identified the circNDRG1 (has_circ_0085656) as a novel candidate circRNA. Transwell and wound-healing assay demonstrated that circNDRG1 inhibited the metastasis of bladder cancer. ChIP assay showed that circNDRG1 was regulated by the transcription factor EGR1 by binding the promotor of host gene NDRG1. RNA-pulldown assay proved that circNDRG1 sponged miR-520h leading to the overexpression of smad7, which was a negative regulatory protein of EMT. Conclusions: Our research revealed that nitroxoline may suppress metastasis in bladder cancer via EGR1/circNDRG1/miR-520h/smad7/EMT signaling pathway.

Journal of Experimental & Clinical Cancer Research

Background Circular RNA (circRNA) is a novel class noncoding RNA (ncRNA) that plays a critical ro... more Background Circular RNA (circRNA) is a novel class noncoding RNA (ncRNA) that plays a critical role in various cancers, including prostate cancer (PCa). However, the clinical significance, biological function, and molecular mechanisms of circRNAs in prostate cancer remain to be elucidated. Methods A circRNA array was performed to identified the differentially expressed circRNAs. circPDE5A was identified as a novel circRNA which downregulated in clinical samples. Functionally, the in vitro and in vivo assays were applied to explore the role of circPDE5A in PCa metastasis. Mechanistically, the interaction between circPDE5A and WTAP was verified using RNA pulldown followed by mass spectrometry, RNA Immunoprecipitation (RIP) assays. m6A methylated RNA immunoprecipitation sequencing (MeRIP-seq) was then used to identified the downstream target of circPDE5A. Chromatin immunoprecipitation assay (ChIP) and dual-luciferase reporter assay were used to identified transcriptional factor which r...

2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021

Trajectory prediction is confronted with the dilemma to capture the multi-modal nature of future ... more Trajectory prediction is confronted with the dilemma to capture the multi-modal nature of future dynamics with both diversity and accuracy. In this paper, we present a distribution discrimination (DisDis) method to predict personalized motion patterns by distinguishing the potential distributions. Motivated by that the motion pattern of each person is personalized due to his/her habit, our DisDis learns the latent distribution to represent different motion patterns and optimize it by the contrastive discrimination. This distribution discrimination encourages latent distributions to be more discriminative. Our method can be integrated with existing multi-modal stochastic predictive models as a plug-and-play module to learn the more discriminative latent distribution. To evaluate the latent distribution, we further propose a new metric, probability cumulative minimum distance (PCMD) curve, which cumulatively calculates the minimum distance on the sorted probabilities. Experimental results on the ETH and UCY datasets show the effectiveness of our method.

Computer Vision – ECCV 2020, 2020

Unlike most existing works that define room layout on a 2D image, we model the layout in 3D as a ... more Unlike most existing works that define room layout on a 2D image, we model the layout in 3D as a configuration of the camera and the room. Our spatial geometric representation with only seven variables is more concise but effective, and more importantly enables direct 3D reasoning, e.g. how the camera is positioned relative to the room. This is particularly valuable in applications such as indoor robot navigation. We formulate the problem as a Markov decision process, in which the layout is incrementally adjusted based on the difference between the current layout and the target image, and the policy is learned via deep reinforcement learning. Our framework is end-to-end trainable, requiring no extra optimization, and achieves competitive performance on two challenging room layout datasets.

Indian Journal of Hematology and Blood Transfusion, 2021

Limited treatment options are available for relapsed or refractory diffuse large B cell lymphoma ... more Limited treatment options are available for relapsed or refractory diffuse large B cell lymphoma (RR DLBCL). Few clinical studies have reported the use of Ibrutinib, a covalent Bruton Tyrosine kinase (BTK) inhibitor, in RR DLBCL. There are relatively few clinical studies about Ibrutinib in RR DLBCL now. We retrospectively investigated the safety and efficacy of Ibrutinib (alone or in combination with other drugs) in patients with RR DLBCL. We reviewed the medical records of 40 RR DLBCL patients who received Ibrutinib alone or in combination with other drugs in our hospital from June 2018 to August 2020. The objective response rate (ORR) of RR DLBCL patients on Ibrutinib was 22.5%. The median progression free survival time (PFS) was 13.0 months (95% CI 8.914–17.086), and the median overall survival time (OS) was 15.0 months (95% CI 11.931–18.089). Rash (25.0%) and fatigue (25.0%) were the most common adverse reactions in this study. The application of Ibrutinib to patients with RR DL...

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019

In this paper, we propose a Enhanced Bayesian Compression method to flexibly compress the deep ne... more In this paper, we propose a Enhanced Bayesian Compression method to flexibly compress the deep networks via reinforcement learning. Unlike existing Bayesian compression methods which can not explicitly enforce quantization weights during training, our method learns flexible codebooks in each layer for an optimal network quantization. To dynamically adjust the state of codebooks, we employ an Actor-Critic network to collaborate with the original deep network. Unlike most existing network quantization methods, our EBC doesn't require retraining procedures after the quantization. Experimental results show that our method obtains low-bit precision with acceptable accuracy drop on MNIST, CIFAR and ImageNet.

SSRN Electronic Journal, 2021

Liver plays a unique role as a metabolic center of the body, and also performs other important fu... more Liver plays a unique role as a metabolic center of the body, and also performs other important functions such as detoxification and immune response. The regulatory networks are formed between hepatocytes and non-parenchymal liver cells in human liver to maintain liver homeostasis. But an in-depth knowledge of the constituent proteins in different human liver cell types is still lacking. Here, we determine the healthy human liver proteome by measuring four major cell types that build classical liver lobule, hepatocytes (HCs), hepatic stellate cells (HSCs), Kupffer cells (KCs), and liver sinusoidal endothelial cells (LSECs) by high-resolution mass spectrometry-based proteomics. Overall, we quantify a total of 8,354 proteins for four cell types and over 6,000 proteins for each cell type. Analysis of this dataset and regulatory pathway reveals the cellular labor division in human liver follows the pattern that parenchymal cells make the main components of pathways, but non-parenchymal c...

In this paper, we propose a multi-agent learning framework to model communication in complex mult... more In this paper, we propose a multi-agent learning framework to model communication in complex multi-agent systems. Most existing multi-agent reinforcement learning methods require agents to exchange information with the environment or global manager to achieve effective and efficient interaction. We model the multi-agent system with an online adaptive graph where all agents communicate with each other through the edges. We update the graph network with a relation system which takes the current graph network and the hidden variable of agents as input. Messages and rewards are shared through the graph network. Finally, we optimize the whole system via the policy gradient algorithm. Experimental results of several multi-agent systems show the efficiency of the proposed method and its strength compared to existing methods in cooperative scenarios.

Umbilical cord blood allogeneic hematopoietic stem cell transplantation(UCBT) has been gradually ... more Umbilical cord blood allogeneic hematopoietic stem cell transplantation(UCBT) has been gradually applied in the treatment of patients with blood system diseases. This paper reports a case of a child patient with highly invasive T-cell lymphoma who underwent UCBT after chemotherapy and developed minimal change glomerulopathy after transplantation.

Computer Vision – ECCV 2018, 2018

In this paper, we propose a simple yet effective relaxationfree method to learn more effective bi... more In this paper, we propose a simple yet effective relaxationfree method to learn more effective binary codes via policy gradient for scalable image search. While a variety of deep hashing methods have been proposed in recent years, most of them are confronted by the dilemma to obtain optimal binary codes in a truly end-to-end manner with nonsmooth sign activations. Unlike existing methods which usually employ a general relaxation framework to adapt to the gradient-based algorithms, our approach formulates the non-smooth part of the hashing network as sampling with a stochastic policy, so that the retrieval performance degradation caused by the relaxation can be avoided. Specifically, our method directly generates the binary codes and maximizes the expectation of rewards for similarity preservation, where the network can be trained directly via policy gradient. Hence, the differentiation challenge for discrete optimization can be naturally addressed, which leads to effective gradients and binary codes. Extensive experimental results on three benchmark datasets validate the effectiveness of the proposed method.

The Journal of Gene Medicine, 2021

Aberrant expression of m6A‐related proteins contributes to the occurrence and progression of non‐... more Aberrant expression of m6A‐related proteins contributes to the occurrence and progression of non‐small cell lung cancer (NSCLC). Current studies mainly focus on single m6A regulatory genes and their underlying mechanisms, and the expression of multiple m6A regulatory proteins in NSCLC remains unclear. Therefore, it is necessary to systematically examine these proteins, particularly in clinical specimens.

Frontiers in Molecular Biosciences, 2021

Accumulating evidence indicates that hypoxia is highly associated with bladder cancer genesis, pr... more Accumulating evidence indicates that hypoxia is highly associated with bladder cancer genesis, progression, and immune microenvironment. Nevertheless, few studies have identified the role of hypoxia-related genes as a prognostic signature in bladder cancer. This study aimed to establish a hypoxia-related signature with high accuracy for prognosis and immune microenvironment prediction in bladder cancer. We obtained expression profiles and clinical information from Gene Expression Omnibus and The Cancer Genome Atlas. Then the univariate Cox regression, random survival forest algorithm, and multivariate Cox regression analysis were conducted to identify the core genes and four hypoxia-related genes (ANXA2, GALK1, COL5A1, and HS3ST1) were selected to construct the signature. Kaplan-Meier survival analysis demonstrated that patients with a low-risk score had a higher disease-specific survival rate (p < 0.0001). The areas under the curve of the signature were 0.829 at 1 year, 0.869 at...

2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017

In this paper, we propose a consistent-aware deep learning (CADL) approach for person re-identifi... more In this paper, we propose a consistent-aware deep learning (CADL) approach for person re-identification in a camera network. Unlike most existing person re-identification methods which identify whether two pedestrian images are from the same person or not, our approach aims to obtain the maximal correct matches for the whole camera network. Different from recently proposed camera network based reidentification methods which only consider the consistent information in the matching stage to obtain a globally optimal association, we exploit such consistent-aware information under a deep learning framework where both feature representation and image matching are automatically learned. Specifically, we reach the globally optimal solution and balance the performance between different cameras by optimizing the similarity and data association iteratively with certain consistent constraints. Experimental results show that our method obtains significant performance improvement and outperforms the state-of-the-art methods by large margins.

Advanced Science, 2020

The incidence of bone metastases in hepatocellular carcinoma (HCC) has increased prominently over... more The incidence of bone metastases in hepatocellular carcinoma (HCC) has increased prominently over the past decade owing to the prolonged overall survival of HCC patients. However, the mechanisms underlying HCC bone‐metastasis remain largely unknown. In the current study, HCC‐secreted lectin galactoside‐binding soluble 3 (LGALS3) is found to be significantly upregulated and correlates with shorter bone‐metastasis‐free survival of HCC patients. Overexpression of LGALS3 enhances the metastatic capability of HCC cells to bone and induces skeletal‐related events by forming a bone pre‐metastatic niche via promoting osteoclast fusion and podosome formation. Mechanically, ubiquitin ligaseRNF219‐meidated α‐catenin degradation prompts YAP1/β‐catenin complex‐dependent epigenetic modifications of LGALS3 promoter, resulting in LGALS3 upregulation and metastatic bone diseases. Importantly, treatment with verteporfin, a clinical drug for macular degeneration, decreases LGALS3 expression and effectively inhibits skeletal complications of HCC. These findings unveil a plausible role for HCC‐secreted LGALS3 in pre‐metastatic niche and can suggest a promising strategy for clinical intervention in HCC bone‐metastasis.

Frontiers in Molecular Biosciences, 2021

Testicular nuclear receptor 4 (TR4) is a member of the nuclear hormone receptor family and acts a... more Testicular nuclear receptor 4 (TR4) is a member of the nuclear hormone receptor family and acts as a ligand-activated transcription factor and functions in many biological processes, such as development, cellular differentiation, and homeostasis. Recent studies have shown that TR4 plays an important role in prostate cancer, renal cell carcinoma, and hepatocellular carcinoma; however, its potential link to bladder cancer (BC) remains unknown. This study found that bladder cancer exhibited a higher expression of TR4 compared to normal tissues. Overexpressed TR4 promoted the bladder cancer cell proliferation, and knocked down TR4 with TR4-siRNA suppressed the bladder cancer cell proliferation. Mechanistic studies reveal that TR4 functions by altering the expression of Bcl-2 to regulate apoptosis in bladder cancer cells. Furthermore, knocking down Bcl-2 reversed the BC proliferation induced by TR4. In vivo, we also confirmed that TR4 knockdown mice (TR4+/−) showed slower bladder cancer ...

Computer Vision – ECCV 2018, 2018

Visual tracking is confronted by the dilemma to locate a target both accurately and efficiently, ... more Visual tracking is confronted by the dilemma to locate a target both accurately and efficiently, and make decisions online whether and how to adapt the appearance model or even restart tracking. In this paper, we propose a deep reinforcement learning with iterative shift (DRL-IS) method for single object tracking, where an actor-critic network is introduced to predict the iterative shifts of object bounding boxes, and evaluate the shifts to take actions on whether to update object models or re-initialize tracking. Since locating an object is achieved by an iterative shift process, rather than online classification on many sampled locations, the proposed method is robust to cope with large deformations and abrupt motion, and computationally efficient since finding a target takes up to 10 shifts. In offline training, the critic network guides to learn how to make decisions jointly on motion estimation and tracking status in an end-to-end manner. Experimental results on the OTB benchmarks with large deformation improve the tracking precision by 1.7% and runs about 5 times faster than the competing state-of-the-art methods.

Molecular Therapy - Oncolytics, 2021

Aberrant expression of the zinc finger protein (ZIC) family has been extensively reported to cont... more Aberrant expression of the zinc finger protein (ZIC) family has been extensively reported to contribute to progression and metastasis in multiple human cancers. However, the functional roles and underlying mechanisms of ZIC2 in non-small cell lung cancer (NSCLC) are largely unknown. In this study, ZIC2 expression was evaluated using qRT-PCR, western blot, and immunohistochemistry, respectively. Animal experiments in vivo and functional assays in vitro were performed to investigate the role of ZIC2 in NSCLC. Luciferase assays and chromatin immunoprecipitation (ChIP) were carried out to explore the underlying target involved in the roles of ZIC2 in NSCLC. Here, we reported that ZIC2 was upregulated in NSCLC tissues, and high expression of ZIC2 predicted worse overall and progression-free survival of NSCLC patients. Silencing ZIC2 repressed tumorigenesis and reduced the anoikis resistance of NSCLC cells. Mechanical investigation further revealed that silencing ZIC2 transcriptionally inhibited Src expression and inactivated steroid receptor coactivator/focal adhesion kinase signaling, which further attenuated the anoikis resistance of NSCLC cells. Importantly, our results showed that the number of circulating tumor cells (CTCs) was positively correlated with ZIC2 expression in NSCLC patients. Collectively, our findings unravel a novel mechanism implicating ZIC2 in NSCLC, which will facilitate the development of anti-tumor strategies in NSCLC.

Cornell University - arXiv, Sep 1, 2022

The point cloud based 3D single object tracking (3DSOT) has drawn increasing attention. Lots of b... more The point cloud based 3D single object tracking (3DSOT) has drawn increasing attention. Lots of breakthroughs have been made, but we also reveal two severe issues. By an extensive analysis, we find the prediction manner of current approaches is non-robust, i.e., exposing a misalignment gap between prediction score and actually localization accuracy. Another issue is the sparse point returns will damage the feature matching procedure of the SOT task. Based on these insights, we introduce two novel modules, i.e., Adaptive Refine Prediction (ARP) and Target Knowledge Transfer (TKT), to tackle them, respectively. To this end, we first design a strong pipeline to extract discriminative features and conduct the matching procedure with the attention mechanism. Then, ARP module is proposed to tackle the misalignment issue by aggregating all predicted candidates with valuable clues. Finally, TKT module is designed to effectively overcome incomplete point cloud due to sparse and occlusion issues. We call our overall framework PCET. By conducting extensive experiments on the KITTI and Waymo Open Dataset, our model achieves state-ofthe-art performance while maintaining a lower computational consumption.

Frontiers in Oncology

The majority of occult liver metastases cannot be detected by computed tomography (CT), magnetic ... more The majority of occult liver metastases cannot be detected by computed tomography (CT), magnetic resonance imaging (MRI) or other traditionally morphological imaging approaches since the lesions are too small or they have not yet formed cancer nodules. Gankyrin is a small molecular protein composed of seven ankyrin domains. In this study, the expression of Gankyrin in colorectal cancer (CRC) patients with liver metastases was investigated to determine its prognosis value. Gankyrin expression in CRC patients was initially analyzed using data from The Cancer Genome Atlas (TCGA) database and bioinformatics tools. RT-qPCR, western blotting, immunohistochemistry (IHC) and transwell migration and invasion assays were then performed to verify the expression and function of Gankyrin in CRC cell line, CRC tissues and matched non-tumor tissues of clinical patients. General clinicopathological information including TNM stage as well as preoperative and postoperative imaging results were collec...

PeerJ

Purpose We aimed to establish a cholesterogenic gene signature to predict the prognosis of young ... more Purpose We aimed to establish a cholesterogenic gene signature to predict the prognosis of young breast cancer (BC) patients and then verified it using cell line experiments. Methods In the bioinformatic section, transcriptional data and corresponding clinical data of young BC patients (age ≤ 45 years) were downloaded from The Cancer Genome Atlas (TCGA) database for training set. Differentially expressed genes (DEGs) were compared between tumour tissue (n = 183) and normal tissue (n = 30). By using univariate Cox regression and multi COX regression, a five-cholesterogenic-gene signature was established to predict prognosis. Subgroup analysis and external validations of GSE131769 from the Gene Expression Omnibus (GEO) were performed to verify the signature. Subsequently, in experiment part, cell experiments were performed to further verify the biological roles of the five cholesterogenic genes in BC. Results In the bioinformatic section, a total of 97 upregulated genes and 124 downre...

International Journal of Biological Sciences

Bladder cancer is one of the most common and deadly cancer worldwide. Current chemotherapy has sh... more Bladder cancer is one of the most common and deadly cancer worldwide. Current chemotherapy has shown limited efficacy in improving outcomes for patients. Nitroxoline, an old and widely used oral antibiotic, which was known to treat for urinary tract infection for decades. Recent studies suggested that nitroxoline suppressed the tumor progression and metastasis, especially in bladder cancer. However, the underlying mechanism for anti-tumor activity of nitroxoline remains unclear. Methods: CircRNA microarray was used to explore the nitroxoline-mediated circRNA expression profile of bladder cancer lines. Transwell and wound-healing assay were applied to evaluate the capacity of metastasis. ChIP assay was chosen to prove the binding of promotor and transcription factor. RNA-pulldown assay was performed to explore the sponge of circRNA and microRNA. Results: We first identified the circNDRG1 (has_circ_0085656) as a novel candidate circRNA. Transwell and wound-healing assay demonstrated that circNDRG1 inhibited the metastasis of bladder cancer. ChIP assay showed that circNDRG1 was regulated by the transcription factor EGR1 by binding the promotor of host gene NDRG1. RNA-pulldown assay proved that circNDRG1 sponged miR-520h leading to the overexpression of smad7, which was a negative regulatory protein of EMT. Conclusions: Our research revealed that nitroxoline may suppress metastasis in bladder cancer via EGR1/circNDRG1/miR-520h/smad7/EMT signaling pathway.

Journal of Experimental & Clinical Cancer Research

Background Circular RNA (circRNA) is a novel class noncoding RNA (ncRNA) that plays a critical ro... more Background Circular RNA (circRNA) is a novel class noncoding RNA (ncRNA) that plays a critical role in various cancers, including prostate cancer (PCa). However, the clinical significance, biological function, and molecular mechanisms of circRNAs in prostate cancer remain to be elucidated. Methods A circRNA array was performed to identified the differentially expressed circRNAs. circPDE5A was identified as a novel circRNA which downregulated in clinical samples. Functionally, the in vitro and in vivo assays were applied to explore the role of circPDE5A in PCa metastasis. Mechanistically, the interaction between circPDE5A and WTAP was verified using RNA pulldown followed by mass spectrometry, RNA Immunoprecipitation (RIP) assays. m6A methylated RNA immunoprecipitation sequencing (MeRIP-seq) was then used to identified the downstream target of circPDE5A. Chromatin immunoprecipitation assay (ChIP) and dual-luciferase reporter assay were used to identified transcriptional factor which r...

2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021

Trajectory prediction is confronted with the dilemma to capture the multi-modal nature of future ... more Trajectory prediction is confronted with the dilemma to capture the multi-modal nature of future dynamics with both diversity and accuracy. In this paper, we present a distribution discrimination (DisDis) method to predict personalized motion patterns by distinguishing the potential distributions. Motivated by that the motion pattern of each person is personalized due to his/her habit, our DisDis learns the latent distribution to represent different motion patterns and optimize it by the contrastive discrimination. This distribution discrimination encourages latent distributions to be more discriminative. Our method can be integrated with existing multi-modal stochastic predictive models as a plug-and-play module to learn the more discriminative latent distribution. To evaluate the latent distribution, we further propose a new metric, probability cumulative minimum distance (PCMD) curve, which cumulatively calculates the minimum distance on the sorted probabilities. Experimental results on the ETH and UCY datasets show the effectiveness of our method.

Computer Vision – ECCV 2020, 2020

Unlike most existing works that define room layout on a 2D image, we model the layout in 3D as a ... more Unlike most existing works that define room layout on a 2D image, we model the layout in 3D as a configuration of the camera and the room. Our spatial geometric representation with only seven variables is more concise but effective, and more importantly enables direct 3D reasoning, e.g. how the camera is positioned relative to the room. This is particularly valuable in applications such as indoor robot navigation. We formulate the problem as a Markov decision process, in which the layout is incrementally adjusted based on the difference between the current layout and the target image, and the policy is learned via deep reinforcement learning. Our framework is end-to-end trainable, requiring no extra optimization, and achieves competitive performance on two challenging room layout datasets.

Indian Journal of Hematology and Blood Transfusion, 2021

Limited treatment options are available for relapsed or refractory diffuse large B cell lymphoma ... more Limited treatment options are available for relapsed or refractory diffuse large B cell lymphoma (RR DLBCL). Few clinical studies have reported the use of Ibrutinib, a covalent Bruton Tyrosine kinase (BTK) inhibitor, in RR DLBCL. There are relatively few clinical studies about Ibrutinib in RR DLBCL now. We retrospectively investigated the safety and efficacy of Ibrutinib (alone or in combination with other drugs) in patients with RR DLBCL. We reviewed the medical records of 40 RR DLBCL patients who received Ibrutinib alone or in combination with other drugs in our hospital from June 2018 to August 2020. The objective response rate (ORR) of RR DLBCL patients on Ibrutinib was 22.5%. The median progression free survival time (PFS) was 13.0 months (95% CI 8.914–17.086), and the median overall survival time (OS) was 15.0 months (95% CI 11.931–18.089). Rash (25.0%) and fatigue (25.0%) were the most common adverse reactions in this study. The application of Ibrutinib to patients with RR DL...

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019

In this paper, we propose a Enhanced Bayesian Compression method to flexibly compress the deep ne... more In this paper, we propose a Enhanced Bayesian Compression method to flexibly compress the deep networks via reinforcement learning. Unlike existing Bayesian compression methods which can not explicitly enforce quantization weights during training, our method learns flexible codebooks in each layer for an optimal network quantization. To dynamically adjust the state of codebooks, we employ an Actor-Critic network to collaborate with the original deep network. Unlike most existing network quantization methods, our EBC doesn't require retraining procedures after the quantization. Experimental results show that our method obtains low-bit precision with acceptable accuracy drop on MNIST, CIFAR and ImageNet.

SSRN Electronic Journal, 2021

Liver plays a unique role as a metabolic center of the body, and also performs other important fu... more Liver plays a unique role as a metabolic center of the body, and also performs other important functions such as detoxification and immune response. The regulatory networks are formed between hepatocytes and non-parenchymal liver cells in human liver to maintain liver homeostasis. But an in-depth knowledge of the constituent proteins in different human liver cell types is still lacking. Here, we determine the healthy human liver proteome by measuring four major cell types that build classical liver lobule, hepatocytes (HCs), hepatic stellate cells (HSCs), Kupffer cells (KCs), and liver sinusoidal endothelial cells (LSECs) by high-resolution mass spectrometry-based proteomics. Overall, we quantify a total of 8,354 proteins for four cell types and over 6,000 proteins for each cell type. Analysis of this dataset and regulatory pathway reveals the cellular labor division in human liver follows the pattern that parenchymal cells make the main components of pathways, but non-parenchymal c...

In this paper, we propose a multi-agent learning framework to model communication in complex mult... more In this paper, we propose a multi-agent learning framework to model communication in complex multi-agent systems. Most existing multi-agent reinforcement learning methods require agents to exchange information with the environment or global manager to achieve effective and efficient interaction. We model the multi-agent system with an online adaptive graph where all agents communicate with each other through the edges. We update the graph network with a relation system which takes the current graph network and the hidden variable of agents as input. Messages and rewards are shared through the graph network. Finally, we optimize the whole system via the policy gradient algorithm. Experimental results of several multi-agent systems show the efficiency of the proposed method and its strength compared to existing methods in cooperative scenarios.

Umbilical cord blood allogeneic hematopoietic stem cell transplantation(UCBT) has been gradually ... more Umbilical cord blood allogeneic hematopoietic stem cell transplantation(UCBT) has been gradually applied in the treatment of patients with blood system diseases. This paper reports a case of a child patient with highly invasive T-cell lymphoma who underwent UCBT after chemotherapy and developed minimal change glomerulopathy after transplantation.

Computer Vision – ECCV 2018, 2018

In this paper, we propose a simple yet effective relaxationfree method to learn more effective bi... more In this paper, we propose a simple yet effective relaxationfree method to learn more effective binary codes via policy gradient for scalable image search. While a variety of deep hashing methods have been proposed in recent years, most of them are confronted by the dilemma to obtain optimal binary codes in a truly end-to-end manner with nonsmooth sign activations. Unlike existing methods which usually employ a general relaxation framework to adapt to the gradient-based algorithms, our approach formulates the non-smooth part of the hashing network as sampling with a stochastic policy, so that the retrieval performance degradation caused by the relaxation can be avoided. Specifically, our method directly generates the binary codes and maximizes the expectation of rewards for similarity preservation, where the network can be trained directly via policy gradient. Hence, the differentiation challenge for discrete optimization can be naturally addressed, which leads to effective gradients and binary codes. Extensive experimental results on three benchmark datasets validate the effectiveness of the proposed method.

The Journal of Gene Medicine, 2021

Aberrant expression of m6A‐related proteins contributes to the occurrence and progression of non‐... more Aberrant expression of m6A‐related proteins contributes to the occurrence and progression of non‐small cell lung cancer (NSCLC). Current studies mainly focus on single m6A regulatory genes and their underlying mechanisms, and the expression of multiple m6A regulatory proteins in NSCLC remains unclear. Therefore, it is necessary to systematically examine these proteins, particularly in clinical specimens.

Frontiers in Molecular Biosciences, 2021

Accumulating evidence indicates that hypoxia is highly associated with bladder cancer genesis, pr... more Accumulating evidence indicates that hypoxia is highly associated with bladder cancer genesis, progression, and immune microenvironment. Nevertheless, few studies have identified the role of hypoxia-related genes as a prognostic signature in bladder cancer. This study aimed to establish a hypoxia-related signature with high accuracy for prognosis and immune microenvironment prediction in bladder cancer. We obtained expression profiles and clinical information from Gene Expression Omnibus and The Cancer Genome Atlas. Then the univariate Cox regression, random survival forest algorithm, and multivariate Cox regression analysis were conducted to identify the core genes and four hypoxia-related genes (ANXA2, GALK1, COL5A1, and HS3ST1) were selected to construct the signature. Kaplan-Meier survival analysis demonstrated that patients with a low-risk score had a higher disease-specific survival rate (p < 0.0001). The areas under the curve of the signature were 0.829 at 1 year, 0.869 at...

2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017

In this paper, we propose a consistent-aware deep learning (CADL) approach for person re-identifi... more In this paper, we propose a consistent-aware deep learning (CADL) approach for person re-identification in a camera network. Unlike most existing person re-identification methods which identify whether two pedestrian images are from the same person or not, our approach aims to obtain the maximal correct matches for the whole camera network. Different from recently proposed camera network based reidentification methods which only consider the consistent information in the matching stage to obtain a globally optimal association, we exploit such consistent-aware information under a deep learning framework where both feature representation and image matching are automatically learned. Specifically, we reach the globally optimal solution and balance the performance between different cameras by optimizing the similarity and data association iteratively with certain consistent constraints. Experimental results show that our method obtains significant performance improvement and outperforms the state-of-the-art methods by large margins.

Advanced Science, 2020

The incidence of bone metastases in hepatocellular carcinoma (HCC) has increased prominently over... more The incidence of bone metastases in hepatocellular carcinoma (HCC) has increased prominently over the past decade owing to the prolonged overall survival of HCC patients. However, the mechanisms underlying HCC bone‐metastasis remain largely unknown. In the current study, HCC‐secreted lectin galactoside‐binding soluble 3 (LGALS3) is found to be significantly upregulated and correlates with shorter bone‐metastasis‐free survival of HCC patients. Overexpression of LGALS3 enhances the metastatic capability of HCC cells to bone and induces skeletal‐related events by forming a bone pre‐metastatic niche via promoting osteoclast fusion and podosome formation. Mechanically, ubiquitin ligaseRNF219‐meidated α‐catenin degradation prompts YAP1/β‐catenin complex‐dependent epigenetic modifications of LGALS3 promoter, resulting in LGALS3 upregulation and metastatic bone diseases. Importantly, treatment with verteporfin, a clinical drug for macular degeneration, decreases LGALS3 expression and effectively inhibits skeletal complications of HCC. These findings unveil a plausible role for HCC‐secreted LGALS3 in pre‐metastatic niche and can suggest a promising strategy for clinical intervention in HCC bone‐metastasis.

Frontiers in Molecular Biosciences, 2021

Testicular nuclear receptor 4 (TR4) is a member of the nuclear hormone receptor family and acts a... more Testicular nuclear receptor 4 (TR4) is a member of the nuclear hormone receptor family and acts as a ligand-activated transcription factor and functions in many biological processes, such as development, cellular differentiation, and homeostasis. Recent studies have shown that TR4 plays an important role in prostate cancer, renal cell carcinoma, and hepatocellular carcinoma; however, its potential link to bladder cancer (BC) remains unknown. This study found that bladder cancer exhibited a higher expression of TR4 compared to normal tissues. Overexpressed TR4 promoted the bladder cancer cell proliferation, and knocked down TR4 with TR4-siRNA suppressed the bladder cancer cell proliferation. Mechanistic studies reveal that TR4 functions by altering the expression of Bcl-2 to regulate apoptosis in bladder cancer cells. Furthermore, knocking down Bcl-2 reversed the BC proliferation induced by TR4. In vivo, we also confirmed that TR4 knockdown mice (TR4+/−) showed slower bladder cancer ...

Computer Vision – ECCV 2018, 2018

Visual tracking is confronted by the dilemma to locate a target both accurately and efficiently, ... more Visual tracking is confronted by the dilemma to locate a target both accurately and efficiently, and make decisions online whether and how to adapt the appearance model or even restart tracking. In this paper, we propose a deep reinforcement learning with iterative shift (DRL-IS) method for single object tracking, where an actor-critic network is introduced to predict the iterative shifts of object bounding boxes, and evaluate the shifts to take actions on whether to update object models or re-initialize tracking. Since locating an object is achieved by an iterative shift process, rather than online classification on many sampled locations, the proposed method is robust to cope with large deformations and abrupt motion, and computationally efficient since finding a target takes up to 10 shifts. In offline training, the critic network guides to learn how to make decisions jointly on motion estimation and tracking status in an end-to-end manner. Experimental results on the OTB benchmarks with large deformation improve the tracking precision by 1.7% and runs about 5 times faster than the competing state-of-the-art methods.

Molecular Therapy - Oncolytics, 2021

Aberrant expression of the zinc finger protein (ZIC) family has been extensively reported to cont... more Aberrant expression of the zinc finger protein (ZIC) family has been extensively reported to contribute to progression and metastasis in multiple human cancers. However, the functional roles and underlying mechanisms of ZIC2 in non-small cell lung cancer (NSCLC) are largely unknown. In this study, ZIC2 expression was evaluated using qRT-PCR, western blot, and immunohistochemistry, respectively. Animal experiments in vivo and functional assays in vitro were performed to investigate the role of ZIC2 in NSCLC. Luciferase assays and chromatin immunoprecipitation (ChIP) were carried out to explore the underlying target involved in the roles of ZIC2 in NSCLC. Here, we reported that ZIC2 was upregulated in NSCLC tissues, and high expression of ZIC2 predicted worse overall and progression-free survival of NSCLC patients. Silencing ZIC2 repressed tumorigenesis and reduced the anoikis resistance of NSCLC cells. Mechanical investigation further revealed that silencing ZIC2 transcriptionally inhibited Src expression and inactivated steroid receptor coactivator/focal adhesion kinase signaling, which further attenuated the anoikis resistance of NSCLC cells. Importantly, our results showed that the number of circulating tumor cells (CTCs) was positively correlated with ZIC2 expression in NSCLC patients. Collectively, our findings unravel a novel mechanism implicating ZIC2 in NSCLC, which will facilitate the development of anti-tumor strategies in NSCLC.