课程文献中心 更多>>

关键词： Machine learning   Multi-temporal change detection   Landsat satellite imagery   Label-noise robust active learning   Multi-date classification   Automatic sample collection  

摘要： Timely and reliable land-use/land-cover (LULC) change dynamic monitoring is the basis of urban understanding and planning. However, either the training sample shortage or the error accumulation in the multi-temporal processing inevitably restricts the monitoring performance. In this paper, to overcome these problems, we present a label-noise robust active learning method, which automatically collects reliable and informative samples from the images and builds a unified classification system with these augmented samples. In more detail, a Bayesian sample collection process that fuses the unsupervised transition information and the multitemporal land-cover information is designed to provide candidate samples with "from-to" labels. A reliability-based multi-classifier active learning method is then proposed to adaptively allocate the more reliable samples to the classes that are difficulty to classify. Finally, a fusion of the multiple multi-date classifications trained by the selected samples is implemented to identify the change type of interest. The dynamic monitoring results for Shanghai, Shenzhen, and Shiyan in China, two megacities with rapid and obvious urbanization and a small city with relatively slow urbanization, indicate that the proposed method achieves a significantly higher accuracy than the current state-of-the art methods. The sample accuracy verified by the high spatial resolution reference maps endorses the applicability of the sample collection, while the reliability-based active learning further ensures the robustness of the proposed method in the label-noise situation. The presented method was tested in two difficult situations (a small training sample case and a training sample set without joint labeling), so that the robustness and accuracy of the approach can be expected to be of a similar or better quality in cases with more training samples. Given its effectiveness and robustness, the proposed method could be widely applied in LULC change dyn

关键词： Ensemble learning   Active learning   Cancer classification   Gene expression data   Fuzzy set   Rough set  

摘要： Background and Objective: Classification of cancer from gene expression data is one of the major research areas in the field of machine learning and medical science. Generally, conventional supervised methods are not able to produce desired classification accuracy due to inadequate training samples present in gene expression data to train the system. Ensemble-based active learning technique in this situation can be effective as it determines few informative samples by all the base classifiers and ensemble the decisions of all the base classifiers to get the most informative samples. Most informative samples are labeled by the subject experts and those are added to the training set, which can improve the classification accuracy. Method: We propose a novel ensemble-based active learning using fuzzy-rough approach for cancer sample classification from microarray gene expression data. The proposed method is able to deal with the uncertainty, overlap and indiscernibility usually present in the subtype classes of the gene expression data and can improve the accuracy of the individual base classifier in presence of limited training samples. Results: The proposed method is validated using eight microarray gene expression datasets. The performance of the proposed method in terms of classification accuracy, precision, recall, F-1-measures and kappa is compared with six other methods. The improvements in accuracy achieved by the proposed method compared to its nearest competitive methods are 2.96%, 9.34%, 0.93%, 3.69%, 7.2% and 4.53% respectively for Colon cancer, Prostate cancer, SRBCT, Ovarian cancer, DLBCL and Central nervous system datasets. Results of the paired t-test justify the statistical relevance of the results in favor of the proposed method for most of the datasets. Conclusion: The proposed method is an effective general purpose ensemble-based active learning adopting the fuzzy-rough concept and therefore can be applied for other classification problem in future.

关键词： Metagenomics   Skin microbiome   Taxonomic profiling   16S sequencing   WGS sequencing  

摘要： The assessment of microbiome biodiversity is the most common application of metagenomics. While 16S sequencing remains standard procedure for taxonomic profiling of metagenomic data, a growing number of studies have clearly demonstrated biases associated with this method. By using Whole Genome Shotgun sequencing (WGS) metagenomics, most of the known restrictions associated with 16S data are alleviated. However, due to the computationally intensive data analyses and higher sequencing costs, WGS based metagenomics remains a less popular option. Selecting the experiment type that provides a comprehensive, yet manageable amount of information is a challenge encountered in many metagenomics studies. In this work, we created a series of artificial bacterial mixes, each with a different distribution of skin-associated microbial species. These mixes were used to estimate the resolution of two different metagenomic experiments - 16S and WGS - and to evaluate several different bioinformatics approaches for taxonomic read classification. In all test cases, WGS approaches provide much more accurate results, in terms of taxa prediction and abundance estimation, in comparison to those of 16S. Furthermore, we demonstrate that a 16S dataset, analysed using different state of the art techniques and reference databases, can produce widely different results. In light of the fact that most forensic metagenomic analysis are still performed using 16S data, our results are especially important.

关键词： Emotional solidarity   social distance   place attachment   contact   festival   World Heritage Site  

摘要： Building on common identity theory and intergroup contact theory, this study sought to further understanding of people-place relationships by developing a holistic theoretical model to scrutinize place attachment as an antecedent of social distance, mediated by emotional solidarity and moderated by frequency of contact between tourists and residents. Visitor data analyzed with SEM revealed that place dependence is a significant predictor of social distance given it affects affinity positively and avoidance negatively, both of which are mediated by the three dimensions of emotional solidarity. Furthermore, the mediated relationships (via emotional solidarity) between place attachment and social distance vary by level of visitors' frequency of interaction with residents. This study expands current theorization by examining the merits of emotional solidarity as an affective link in a tourist cognitive-behavioral model. From a practical standpoint, DMOs need to understand these construct linkages and include residents in their marketing strategies to increase repeat visitation.

关键词： Hyperspectral imaging   Deoxynivalenol   Near infrared   Cereal sorting   Contamination prediction  

摘要： Near infrared hyperspectral imaging (HSI-NIR) is considered a promising technique able to replace time-consuming, costly and destructive classic methods to predict and classify deoxynivalenol (DON) contaminated wheat kernels or samples by its concentration and level of contamination, respectively. The main objective of the present study was to standardise the HSI-NIR image acquisition method in naturally contaminated whole wheat kernels to obtain a high accuracy method to quantify and classify samples according to DON levels. To confirm the results, wheat samples were analysed by high performance liquid chromatography as the reference method to determine their DON levels. Hyperspectral images for single kernels and whole samples were obtained and spectral data were processed by multivariate analysis software. The initial work revealed that HSI-NIR was able to overcome kernel orientation, position and pixel selection. The subsequent developed Partial Least Squares (PIS) prediction achieved a RMSEP (Root Mean Square Error of Prediction) of 405 mu g/kg and 1174 mu g/kg for a cross-validated model and an independent set validated model, respectively. Moreover, the classification accuracy obtained by Linear Discriminant Analysis (LDA) was 62.7% for two categories depending on the EU maximum level (1250 mu g/kg). Despite of the results are not accurate enough for DON quantification and sample classification, they can be considered a starting point for further improved protocols for DON management.

关键词： Validation   Minisequencing   SNaPshot   Biogeographic origin   Highly degraded samples   Forensic laboratories  

摘要： Mitochondrial DNA (mtDNA) is a useful tool in forensic investigation as it provides information about the matrilineal ancestry of individuals. In addition, mtDNA can be analyzed when the analysis of other nuclear markers is underperforming. Recently, we developed a minisequencing panel for the simultaneous analysis of 52 mtDNA SNPs to classify maternal lineages into the main haplogroups and their phylogeographic origin. In order to make this panel suitable for forensic genetics laboratories, a validation study has been performed in accordance with the Scientific Working Group on DNA Analysis Methods (SWGDAM) guidelines, including species specificity, reproducibility, sensitivity, and stability tests. The results demonstrate that the panel of 52 mtDNA SNPs is highly sensitive, since it enables to obtain complete genetic profiles of samples containing minimal amounts of DNA (1 pg). Furthermore, it provides sufficient genetic information to detect the matrilineal biogeographical origin of highly degraded samples, i.e., ancient dating skeletal remains, and samples with the presence of inhibitors, such as hematin and humic acid. In addition, this panel can detect mixtures in samples whose mtDNA haplogroups of contributors are different. Overall, the results of this study demonstrate the suitability of this minisequencing panel of 52 mtDNA SNPs to be used in forensic cases, with samples of low amount or degraded DNA.