Qing He - Academia.edu (original) (raw)

Papers by Qing He

2012 9th International Conference on Fuzzy Systems and Knowledge Discovery, 2012

Most of the previous works for web video topic detection(e.g., graph-based co-clustering method) ... more Most of the previous works for web video topic detection(e.g., graph-based co-clustering method) always encounter the problem of real-time topic detection, since they all suffer from the high computation complexity. Therefore, a fast topic detection is needed to meet users' or administrators' requirement in real-world scenarios. Along this line, we propose a fast and effective topic detection framework, in which

With rapid growth of the Internet, how to get information from this huge information space become... more With rapid growth of the Internet, how to get information from this huge information space becomes an even important problem. In this paper, An

A universal classification method called Hyper Surface Classifier (HSC) has recently been propose... more A universal classification method called Hyper Surface Classifier (HSC) has recently been proposed. How to sampling the training examples based on HSC is studied in this paper. The concept of Minimum Sample Set is put forward to enhance HSC performance and analyses its generality.

The thermally-induced martensitic transformation in a polycrystalline CuZnAl shape memory alloy (... more The thermally-induced martensitic transformation in a polycrystalline CuZnAl shape memory alloy (SMA) was probed using photostimulated electron emission (PSE) and ultra-violet photoelectron spectroscopy (UPS). UPS measurements of the apparent surface work function during heating show distinct differences that we attribute to electronic structure changes associated with the transformation. In situ photoelectron emission intensities during heating also show distinctive changes that

Clinical cancer research : an official journal of the American Association for Cancer Research, Jan 16, 2015

Chemoresistance is a major obstacle in cancer therapy. We found that fluorouracil (5-FU)-resistan... more Chemoresistance is a major obstacle in cancer therapy. We found that fluorouracil (5-FU)-resistant esophageal squamous cell carcinoma cell lines, established through exposure to increasing concentrations of 5-FU, showed upregulation of Id1, IGF2, and E2F1. We hypothesized that these genes may play an important role in cancer chemoresistance. In vitro and in vivo functional assays were performed to study the effects of Id1-E2F1-IGF2 signaling in chemoresistance. Quantitative real-time PCR, Western blot, immunoprecipitation, chromatin immunoprecipitation, and dual-luciferase reporter assays were used to investigate the molecular mechanisms by which Id1 regulates E2F1 and by which E2F1 regulates IGF2. Clinical specimens, tumor tissue microarray and Gene Expression Omnibus datasets were used to analyze the correlations between gene expressions, and the relationships between expression profiles and patient survival outcomes. Id1 conferred 5-FU chemoresistance through E2F1-dependent induc...

Scientific Reports, 2015

Predicting and understanding the cation distribution in spinels has been one of the most interest... more Predicting and understanding the cation distribution in spinels has been one of the most interesting problems in materials science. The present work investigates the effect of cation redistribution on the structural, electrical, optical and magnetic properties of mixed-valent inverse spinel NiCo2O4(NCO) thin films. It is observed that the films grown at low temperatures (T < 400 °C) exhibit metallic behavior while that grown at higher temperatures (T > 400 °C) are insulators with lower ferrimagnetic-paramagnetic phase transition temperature. So far, n-type Fe3O4 has been used as a conducting layer for the spinel thin films based devices and the search for a p-type counterpart still remains elusive. The inherent coexistence and coupling of ferrimagnetic order and the metallic nature in p-type NCO makes it a promising candidate for spintronic devices. Detailed X-ray Absorption and X-ray Magnetic Circular Dichroism studies revealed a strong correlation between the mixed-valent cation distribution and the resulting ferrimagnetic-metallic/insulating behavior. Our study clearly demonstrates that it is the concentration of Ni(3+)ions and the Ni(3+)-O(2-)Ni(2+) double exchange interaction that is crucial in dictating the metallic behavior in NCO ferrimagnet. The metal-insulator and the associated magnetic order-disorder transitions can be tuned by the degree of cation site disorder via growth conditions.

Based on Jordan Curve Theorem, a universal classification method called HyperSurface Classifier (... more Based on Jordan Curve Theorem, a universal classification method called HyperSurface Classifier (HSC) has recently been proposed. Experimental results show that in three-dimensional space, this method works fairly well in both accuracy and efficiency even for large size data up to 10 7 . However, what we really need is an algorithm that can deal with data not only of massive size but also of high dimensionality. In this paper, an approach based on the idea of classifiers ensemble by dimension dividing without dimension reduction for high dimensional data is proposed. The most important difference between HSC ensemble and the traditional ensemble is that the sub-datasets are obtained by dividing the features rather than by dividing the sample set. Experimental results show that this method has a preferable performance on high dimensional datasets.

International Journal of Pattern Recognition and Artificial Intelligence

Lecture Notes in Computer Science, 2010

Document summarization has attracted a lot of research interest since the 1960s. However, it stil... more Document summarization has attracted a lot of research interest since the 1960s. However, it still remains a challenging task on how to extract effective feature for automatic summarization. In this paper, we extract two features called entropy and relevance to leverage information from different perspectives for summarization. Experiments on unsupervised and supervised methods testify the effectiveness of leveraging the two features.

Neurocomputing, 2015

ABSTRACT Extreme learning machine (ELM) as an emerging technology has achieved exceptional perfor... more ABSTRACT Extreme learning machine (ELM) as an emerging technology has achieved exceptional performance in large-scale settings, and is well suited to binary and multi-class classification, as well as regression tasks. However, existing ELM and its variants predominantly employ single hidden layer feedforward networks, leaving the popular and potentially powerful stacked generalization principle unexploited for seeking predictive deep representations of input data. Deep architectures can find higher-level representations, thus can potentially capture relevant higher-level abstractions. But most of current deep learning methods require solving a difficult and non-convex optimization problem. In this paper, we propose a stacked model, DrELM, to learn deep representations via extreme learning machine according to stacked generalization philosophy. The proposed model utilizes ELM as a base building block and incorporates random shift and kernelization as stacking elements. Specifically, in each layer, DrELM integrates a random projection of the predictions obtained by ELM into the original feature, and then applies kernel functions to generate the resultant feature. To verify the classification and regression performance of DrELM, we conduct the experiments on both synthetic and real-world data sets. The experimental results show that DrELM outperforms ELM and kernel ELMs, which appear to demonstrate that DrELM could yield predictive features that are suitable for prediction tasks. The performances of the deep models (i.e. Stacked Auto-encoder) are comparable. However, due to the utilization of ELM, DrELM is easier to learn and faster in testing.

Supervised dimensionality reduction has shown great advantages in finding predictive subspaces. P... more Supervised dimensionality reduction has shown great advantages in finding predictive subspaces. Previous methods rarely consider the popular maximum margin principle and are prone to overfitting to usually small training data, especially for those under the maximum likelihood framework. In this paper, we present a posterior-regularized Bayesian approach to combine Principal Component Analysis (PCA) with the max-margin learning. Based on the data augmentation idea for max-margin learning and the probabilistic interpretation of PCA, our method can automatically infer the weight and penalty parameter of max-margin learning machine, while finding the most appropriate PCA sub-space simultaneously under the Bayesian framework. We develop a fast mean-field variational inference algorithm to approximate the posterior. Experimental results on various classification tasks show that our method outperforms a number of competitors.

Proceedings of the 19th ACM international conference on Information and knowledge management - CIKM '10, 2010

The distribution difference among multiple data domains has been considered for the cross-domain ... more The distribution difference among multiple data domains has been considered for the cross-domain text classification problem. In this study, we show two new observations along this line. First, the data distribution difference may come from the fact that different domains use different key words to express the same concept. Second, the association between this conceptual feature and the document class may be stable across domains. These two issues are actually the distinction and commonality across data domains.

2010 International Conference on Machine Learning and Cybernetics, 2010

ABSTRACT Support Vector Machine (SVM) is a classification technique of machine learning based on ... more ABSTRACT Support Vector Machine (SVM) is a classification technique of machine learning based on statistical learning theory. A quadratic optimization problem needs to be solved in the algorithm, and with the increase of the samples, the time complexity will also increase. So it is necessary to shrink training sets to reduce the time complexity. A sample selection method for SVM is proposed in this paper. It is inspired from the Hyper surface classification (HSC), which is a universal classification method based on Jordan Curve Theorem, and there is no need for mapping from lower-dimensional space to higher-dimensional space. The experiments show that the algorithm shrinks training sets keeping the accuracy for unseen vectors high.

Photosynthesis research, Jan 13, 2015

CP43 is a chlorophyll-binding protein, which acts as a conduit for the excitation energy transfer... more CP43 is a chlorophyll-binding protein, which acts as a conduit for the excitation energy transfer. The thermal stability of apo-CP43 was studied by intrinsic fluorescence, exogenous ANS fluorescence, and circular dichroism spectroscopy. Under heat treatment, the structure of apo-CP43 changed and existed transition state occurred between 56 and 62 °C by the intrinsic, exogenous ANS fluorescence and the analysis of hydrophobicity. Besides, the isosbestic point of the sigmoidal curve was 58.10 ± 1.02 °C by calculating α-helix transition and the Tm was 56.45 ± 0.52 and 55.59 ± 0.68 °C by calculating the unfolded fraction of tryptophan and tyrosine fluorescence, respectively. During the process of unfolding, the hydrophobic structure of C-terminal segment firstly started to expose at 40 °C, and then the hydrophobic cluster adjacent to the N-terminal segment also gradually exposed to hydrophilic environment with increasing temperature. Our results indicated that heat treatment, especially...

2007 International Conference on Machine Learning and Cybernetics, 2007

For Hyper Surface Classification (HSC), based on the concept of Minimal Consistent Subset for a d... more For Hyper Surface Classification (HSC), based on the concept of Minimal Consistent Subset for a disjoint Cover set (MCSC), a judgmental sampling method is proposed to select a representative subset from the original sample set in this paper. The sampling method depends on sample distribution. HSC can directly solve the nonlinear multi-class classification problems and observe the sample distribution. The sample distribution is obtained by adaptively dividing the sample space, and the classification model of hyper surface is directly used to classify large database based on Jordan Curve Theorem in Topology while sampling for MCSC. The number of MCSC is calculated. MCSC has the same classification model with the entire sample set and can totally reflect its classification ability. For any subset of the sample set that contains MCSC, the classification ability remains the same. Moreover, a formula is put forward that can predict the testing accuracy exactly when some samples are deleted from MCSC. So MCSC is the best way of sampling from the original sample set for Hyper Surface Classification method.

Remote Sensing and Modeling of Ecosystems for Sustainability, 2004

To sum up, the conclusions about the characteristics of the sand-surface reflectivities of visibl... more To sum up, the conclusions about the characteristics of the sand-surface reflectivities of visible light channel and near infrared channel in the Tazhong region in the hinterland of the Taklimakan Desert in winter are as following:(1) On fine days, the sand-surface average reflectivities of visible light channel and near infrared channel in the Tazhong region are 25.8% and 28.6% respectively.

Ecosystems Dynamics, Ecosystem-Society Interactions, and Remote Sensing Applications for Semi-Arid and Arid Land, 2003

After analyzing the heat conditions in the years of serious reduction of cotton yield in the main... more After analyzing the heat conditions in the years of serious reduction of cotton yield in the main cotton-growing areas of Xinjiang, it is found that the cold disasters, especially the delaying cold disasters, are the most serious meteorological disasters to the cotton production in Xinjiang.

CPEM 2010, 2010

Ohmic contacts to 2DEG are crucial in quantized Hall resistance device fabrication. Magnetic prop... more Ohmic contacts to 2DEG are crucial in quantized Hall resistance device fabrication. Magnetic property of the commonly used Au/Ge/Ni recipe for ohmic contact to 2DEG in a GaAs/AlGaAs heterostructure is studied in this summary paper.

CPEM 2010, 2010

Till now, quantized Hall devices distributed by the BIPM are used in NIM's quantized Hall re... more Till now, quantized Hall devices distributed by the BIPM are used in NIM's quantized Hall resistance standard. In this paper, we report the preliminary results of quantized Hall devices with GaAs-AlGaAs heterostructures fabricated by ourselves. The device is with AuGeNi contacts for reliability consideration, but the contacts are with relatively large resistances for layer contents and annealing condition was not optimized for 2DEG and low temperature. The large contact resistance caused additional noises and we observed anomalous values for i = 2 quantized Hall step. Next step, The AuGeNi contacts will be optimized.

2012 9th International Conference on Fuzzy Systems and Knowledge Discovery, 2012

Most of the previous works for web video topic detection(e.g., graph-based co-clustering method) ... more Most of the previous works for web video topic detection(e.g., graph-based co-clustering method) always encounter the problem of real-time topic detection, since they all suffer from the high computation complexity. Therefore, a fast topic detection is needed to meet users' or administrators' requirement in real-world scenarios. Along this line, we propose a fast and effective topic detection framework, in which

With rapid growth of the Internet, how to get information from this huge information space become... more With rapid growth of the Internet, how to get information from this huge information space becomes an even important problem. In this paper, An

A universal classification method called Hyper Surface Classifier (HSC) has recently been propose... more A universal classification method called Hyper Surface Classifier (HSC) has recently been proposed. How to sampling the training examples based on HSC is studied in this paper. The concept of Minimum Sample Set is put forward to enhance HSC performance and analyses its generality.

The thermally-induced martensitic transformation in a polycrystalline CuZnAl shape memory alloy (... more The thermally-induced martensitic transformation in a polycrystalline CuZnAl shape memory alloy (SMA) was probed using photostimulated electron emission (PSE) and ultra-violet photoelectron spectroscopy (UPS). UPS measurements of the apparent surface work function during heating show distinct differences that we attribute to electronic structure changes associated with the transformation. In situ photoelectron emission intensities during heating also show distinctive changes that

Clinical cancer research : an official journal of the American Association for Cancer Research, Jan 16, 2015

Chemoresistance is a major obstacle in cancer therapy. We found that fluorouracil (5-FU)-resistan... more Chemoresistance is a major obstacle in cancer therapy. We found that fluorouracil (5-FU)-resistant esophageal squamous cell carcinoma cell lines, established through exposure to increasing concentrations of 5-FU, showed upregulation of Id1, IGF2, and E2F1. We hypothesized that these genes may play an important role in cancer chemoresistance. In vitro and in vivo functional assays were performed to study the effects of Id1-E2F1-IGF2 signaling in chemoresistance. Quantitative real-time PCR, Western blot, immunoprecipitation, chromatin immunoprecipitation, and dual-luciferase reporter assays were used to investigate the molecular mechanisms by which Id1 regulates E2F1 and by which E2F1 regulates IGF2. Clinical specimens, tumor tissue microarray and Gene Expression Omnibus datasets were used to analyze the correlations between gene expressions, and the relationships between expression profiles and patient survival outcomes. Id1 conferred 5-FU chemoresistance through E2F1-dependent induc...

Scientific Reports, 2015

Predicting and understanding the cation distribution in spinels has been one of the most interest... more Predicting and understanding the cation distribution in spinels has been one of the most interesting problems in materials science. The present work investigates the effect of cation redistribution on the structural, electrical, optical and magnetic properties of mixed-valent inverse spinel NiCo2O4(NCO) thin films. It is observed that the films grown at low temperatures (T < 400 °C) exhibit metallic behavior while that grown at higher temperatures (T > 400 °C) are insulators with lower ferrimagnetic-paramagnetic phase transition temperature. So far, n-type Fe3O4 has been used as a conducting layer for the spinel thin films based devices and the search for a p-type counterpart still remains elusive. The inherent coexistence and coupling of ferrimagnetic order and the metallic nature in p-type NCO makes it a promising candidate for spintronic devices. Detailed X-ray Absorption and X-ray Magnetic Circular Dichroism studies revealed a strong correlation between the mixed-valent cation distribution and the resulting ferrimagnetic-metallic/insulating behavior. Our study clearly demonstrates that it is the concentration of Ni(3+)ions and the Ni(3+)-O(2-)Ni(2+) double exchange interaction that is crucial in dictating the metallic behavior in NCO ferrimagnet. The metal-insulator and the associated magnetic order-disorder transitions can be tuned by the degree of cation site disorder via growth conditions.

Based on Jordan Curve Theorem, a universal classification method called HyperSurface Classifier (... more Based on Jordan Curve Theorem, a universal classification method called HyperSurface Classifier (HSC) has recently been proposed. Experimental results show that in three-dimensional space, this method works fairly well in both accuracy and efficiency even for large size data up to 10 7 . However, what we really need is an algorithm that can deal with data not only of massive size but also of high dimensionality. In this paper, an approach based on the idea of classifiers ensemble by dimension dividing without dimension reduction for high dimensional data is proposed. The most important difference between HSC ensemble and the traditional ensemble is that the sub-datasets are obtained by dividing the features rather than by dividing the sample set. Experimental results show that this method has a preferable performance on high dimensional datasets.

International Journal of Pattern Recognition and Artificial Intelligence

Lecture Notes in Computer Science, 2010

Document summarization has attracted a lot of research interest since the 1960s. However, it stil... more Document summarization has attracted a lot of research interest since the 1960s. However, it still remains a challenging task on how to extract effective feature for automatic summarization. In this paper, we extract two features called entropy and relevance to leverage information from different perspectives for summarization. Experiments on unsupervised and supervised methods testify the effectiveness of leveraging the two features.

Neurocomputing, 2015

ABSTRACT Extreme learning machine (ELM) as an emerging technology has achieved exceptional perfor... more ABSTRACT Extreme learning machine (ELM) as an emerging technology has achieved exceptional performance in large-scale settings, and is well suited to binary and multi-class classification, as well as regression tasks. However, existing ELM and its variants predominantly employ single hidden layer feedforward networks, leaving the popular and potentially powerful stacked generalization principle unexploited for seeking predictive deep representations of input data. Deep architectures can find higher-level representations, thus can potentially capture relevant higher-level abstractions. But most of current deep learning methods require solving a difficult and non-convex optimization problem. In this paper, we propose a stacked model, DrELM, to learn deep representations via extreme learning machine according to stacked generalization philosophy. The proposed model utilizes ELM as a base building block and incorporates random shift and kernelization as stacking elements. Specifically, in each layer, DrELM integrates a random projection of the predictions obtained by ELM into the original feature, and then applies kernel functions to generate the resultant feature. To verify the classification and regression performance of DrELM, we conduct the experiments on both synthetic and real-world data sets. The experimental results show that DrELM outperforms ELM and kernel ELMs, which appear to demonstrate that DrELM could yield predictive features that are suitable for prediction tasks. The performances of the deep models (i.e. Stacked Auto-encoder) are comparable. However, due to the utilization of ELM, DrELM is easier to learn and faster in testing.

Supervised dimensionality reduction has shown great advantages in finding predictive subspaces. P... more Supervised dimensionality reduction has shown great advantages in finding predictive subspaces. Previous methods rarely consider the popular maximum margin principle and are prone to overfitting to usually small training data, especially for those under the maximum likelihood framework. In this paper, we present a posterior-regularized Bayesian approach to combine Principal Component Analysis (PCA) with the max-margin learning. Based on the data augmentation idea for max-margin learning and the probabilistic interpretation of PCA, our method can automatically infer the weight and penalty parameter of max-margin learning machine, while finding the most appropriate PCA sub-space simultaneously under the Bayesian framework. We develop a fast mean-field variational inference algorithm to approximate the posterior. Experimental results on various classification tasks show that our method outperforms a number of competitors.

Proceedings of the 19th ACM international conference on Information and knowledge management - CIKM '10, 2010

The distribution difference among multiple data domains has been considered for the cross-domain ... more The distribution difference among multiple data domains has been considered for the cross-domain text classification problem. In this study, we show two new observations along this line. First, the data distribution difference may come from the fact that different domains use different key words to express the same concept. Second, the association between this conceptual feature and the document class may be stable across domains. These two issues are actually the distinction and commonality across data domains.

2010 International Conference on Machine Learning and Cybernetics, 2010

ABSTRACT Support Vector Machine (SVM) is a classification technique of machine learning based on ... more ABSTRACT Support Vector Machine (SVM) is a classification technique of machine learning based on statistical learning theory. A quadratic optimization problem needs to be solved in the algorithm, and with the increase of the samples, the time complexity will also increase. So it is necessary to shrink training sets to reduce the time complexity. A sample selection method for SVM is proposed in this paper. It is inspired from the Hyper surface classification (HSC), which is a universal classification method based on Jordan Curve Theorem, and there is no need for mapping from lower-dimensional space to higher-dimensional space. The experiments show that the algorithm shrinks training sets keeping the accuracy for unseen vectors high.

Photosynthesis research, Jan 13, 2015

CP43 is a chlorophyll-binding protein, which acts as a conduit for the excitation energy transfer... more CP43 is a chlorophyll-binding protein, which acts as a conduit for the excitation energy transfer. The thermal stability of apo-CP43 was studied by intrinsic fluorescence, exogenous ANS fluorescence, and circular dichroism spectroscopy. Under heat treatment, the structure of apo-CP43 changed and existed transition state occurred between 56 and 62 °C by the intrinsic, exogenous ANS fluorescence and the analysis of hydrophobicity. Besides, the isosbestic point of the sigmoidal curve was 58.10 ± 1.02 °C by calculating α-helix transition and the Tm was 56.45 ± 0.52 and 55.59 ± 0.68 °C by calculating the unfolded fraction of tryptophan and tyrosine fluorescence, respectively. During the process of unfolding, the hydrophobic structure of C-terminal segment firstly started to expose at 40 °C, and then the hydrophobic cluster adjacent to the N-terminal segment also gradually exposed to hydrophilic environment with increasing temperature. Our results indicated that heat treatment, especially...

2007 International Conference on Machine Learning and Cybernetics, 2007

For Hyper Surface Classification (HSC), based on the concept of Minimal Consistent Subset for a d... more For Hyper Surface Classification (HSC), based on the concept of Minimal Consistent Subset for a disjoint Cover set (MCSC), a judgmental sampling method is proposed to select a representative subset from the original sample set in this paper. The sampling method depends on sample distribution. HSC can directly solve the nonlinear multi-class classification problems and observe the sample distribution. The sample distribution is obtained by adaptively dividing the sample space, and the classification model of hyper surface is directly used to classify large database based on Jordan Curve Theorem in Topology while sampling for MCSC. The number of MCSC is calculated. MCSC has the same classification model with the entire sample set and can totally reflect its classification ability. For any subset of the sample set that contains MCSC, the classification ability remains the same. Moreover, a formula is put forward that can predict the testing accuracy exactly when some samples are deleted from MCSC. So MCSC is the best way of sampling from the original sample set for Hyper Surface Classification method.

Remote Sensing and Modeling of Ecosystems for Sustainability, 2004

To sum up, the conclusions about the characteristics of the sand-surface reflectivities of visibl... more To sum up, the conclusions about the characteristics of the sand-surface reflectivities of visible light channel and near infrared channel in the Tazhong region in the hinterland of the Taklimakan Desert in winter are as following:(1) On fine days, the sand-surface average reflectivities of visible light channel and near infrared channel in the Tazhong region are 25.8% and 28.6% respectively.

Ecosystems Dynamics, Ecosystem-Society Interactions, and Remote Sensing Applications for Semi-Arid and Arid Land, 2003

After analyzing the heat conditions in the years of serious reduction of cotton yield in the main... more After analyzing the heat conditions in the years of serious reduction of cotton yield in the main cotton-growing areas of Xinjiang, it is found that the cold disasters, especially the delaying cold disasters, are the most serious meteorological disasters to the cotton production in Xinjiang.

CPEM 2010, 2010

Ohmic contacts to 2DEG are crucial in quantized Hall resistance device fabrication. Magnetic prop... more Ohmic contacts to 2DEG are crucial in quantized Hall resistance device fabrication. Magnetic property of the commonly used Au/Ge/Ni recipe for ohmic contact to 2DEG in a GaAs/AlGaAs heterostructure is studied in this summary paper.

CPEM 2010, 2010

Till now, quantized Hall devices distributed by the BIPM are used in NIM's quantized Hall re... more Till now, quantized Hall devices distributed by the BIPM are used in NIM's quantized Hall resistance standard. In this paper, we report the preliminary results of quantized Hall devices with GaAs-AlGaAs heterostructures fabricated by ourselves. The device is with AuGeNi contacts for reliability consideration, but the contacts are with relatively large resistances for layer contents and annealing condition was not optimized for 2DEG and low temperature. The large contact resistance caused additional noises and we observed anomalous values for i = 2 quantized Hall step. Next step, The AuGeNi contacts will be optimized.