Multi-label learning is a class of machine learning algorithms that study the classification problem of data associated with multiple labels simultaneously. Ensemble-based method is one of the ...representative methods in multi-label learning. In the existing ensemble-based multi-label classification methods, the essence of its base learner is still a binary or multi-class classifier. At present, there is no ensemble-based multi-label method that uses a multi-label classifier based on algorithm adaptation as a base learner. Multi-label algorithms suffer from expensive time costs, and ensemble-based methods further exacerbate the time cost. To address these issues, we propose an efficient multi-label classification method based on kernel extreme learning machine and ensemble learning. Firstly, to solve the problem of high time complexity of multi-label classifiers, we use a Gaussian kernel function to generate a kernel extreme learning machine based multi-label classifier (KELM_ML). Then, based on random sampling, we construct the base classifier, i.e., KELM_ML, in ensemble learning. Finally, to build an ensemble model of multi-label classifiers, an elimination optimization ensemble strategy is proposed by defining an encoding vector of the base classifiers and updating it. Although our method is an ensemble model, it inherits the advantages of good time complexity of ELM, i.e., fast learning time. The experimental and statistical results show that compared with baseline methods and other ensemble-based multi-label methods, our proposed method has better classification performance and stable results.
Prospective memory (PM) impairment is associated with impaired social functioning, but evidence is limited to chronic schizophrenia samples and cross-sectional design. The aim of this study was to ...utilize network analysis to address the complex interplay between PM, psychopathology, and functional outcome.
This longitudinal study recruited 119 people with first-episode
schizophrenia and followed up with them for 2 to 6 years. PM and working memory were assessed at baseline (in 2010-2015) using valid computerized tasks and the Letter-Number Span Test, respectively. Psychopathology and social functioning were assessed at endpoint (in 2016-2017) using the Positive and Negative Syndrome Scale (PANSS) and the Social and Occupational Functioning Assessment Scale (SOFAS), respectively. Network analysis examined the effect of baseline PM on SOFAS while accounting for the effects of psychopathology.
The resultant network showed that social functioning, PANSS positive symptoms, and PANSS general symptoms clustered together, whereas time-based and event-based PM and working memory formed another cluster. Time-based PM linked event-based PM and working memory with social functioning. Time-based PM (expected influence EI = 0.69), event-based PM (EI = 0.65), and working memory (EI = 0.83) demonstrated high values of expected influence, but social functioning (variance explained = 0.685) and PANSS negative (variance explained = 0.657) and general (variance explained = 0.583) subscales demonstrated high values of predictability.
Time-based PM is the central node linking neurocognitive functions with social functioning. PM and working memory are "target" nodes for interventions bringing changes to the network, whereas social functioning and psychopathology are "malleable" nodes. PM and working memory are promising intervention targets for functional recovery in schizophrenia.
Fuzzy rough sets are the generalization of traditional rough sets to deal with both fuzziness and vagueness in data. The existing researches on fuzzy rough sets are mainly concentrated on the ...construction of approximation operators. Less effort has been put on the attributes reduction of databases with fuzzy rough sets. This paper mainly focuses on the attributes reduction with fuzzy rough sets. After analyzing the previous works on attributes reduction with fuzzy rough sets, we introduce formal concepts of attributes reduction with fuzzy rough sets and completely study the structure of attributes reduction. An algorithm using discernibility matrix to compute all the attributes reductions is developed. Based on these lines of thought, we set up a solid mathematical foundation for attributes reduction with fuzzy rough sets. The experimental results show that the idea in this paper is feasible and valid.
The construction of fuzzy relations is a key issue of fuzzy rough sets. The fuzzy relations generated by the soft distances between samples are more robust than that generated by the hard distances ...between samples. To improve the ability of fuzzy rough sets in deleting redundant attributes, we propose two enhanced fuzzy similarity relations by fully mining neighborhood information and decision information of samples. Then, we establish the Neighborhood Constrained Fuzzy Rough Sets (NC-FRS) by using the proposed relations to perform attribute reduction. Meanwhile, we design enhanced fuzzy similarity relation-based attribute reduction (EFSR-AR) to select important attributes for classification tasks. Finally, we download three gene expression profiles from NCBI to verify that the proposed algorithm can select genes highly related to tumors, the selected genes are more conducive to tumor classification, and the proposed algorithm has strong anti-noise ability. The comparison results indicate that EFSR-AR does have the ability to combat noise and select some genes highly related to tumors.
•A soft distance is constructed by combining neighborhood and decision information.•Proposing a fuzzy relation to enhance generalization ability of rough sets (RS).•Using proposed relation to estimate attributes and perform attribute reduction.•Proposed algorithm can select attributes highly related to classification tasks.•This study deepens the theory and application of RS from perspective of antinoise.
Extreme Learning Machine (ELM) can be considered a probabilistic model, wherein the output is a random variable characterized by its mean and variance, both crucial for assessing and enhancing ELM’s ...generalization ability. Despite its importance, limited research has been conducted on quantitative variance estimation and the underlying factors that influence connecting weight regularization. In this paper, we establish a connection between the variance and weights of ELM for regression problems, and propose maximum likelihood-based estimations for model parameters. We introduce a novel activation function and algorithms to generate hidden layer outputs and predicted outputs that adhere or approximate a normal distribution, as validated through experiments. Additionally, by leveraging the established relationship, we derive estimators for weights, variance, and confidence intervals of predicted values using Maximum Likelihood Estimation (MLE). The variance estimator enables a quantitative evaluation of generalization, and the link between variance and weights offers a theoretical basis for ELM regularization techniques.
Extreme learning machine (ELM) is a non-iterative algorithm for training single-hidden layer feed-forward networks, whose training speed is much faster than those of conventional neural networks. ...However, its objective is only to minimize the empirical risk, which may cause overfitting easily. To overcome this defect, this paper proposes a novel algorithm named LGE2LM based on the localized generalization error model which provides an upper bound of generalization error through adopting the stochastic sensitivity. LGE2LM aims to improve the generalization capability of ELM by using the regularization technique which makes a trade-off between the empirical risk and the stochastic sensitive measure. The essence of LGE2LM is a quadratic problem without iterative process. Similar to ELM, all the parameters of this new algorithm are obtained without tuning, which makes the efficiency of LGE2LM much higher than those of traditional neural networks. Several experiments conducted on both artificial and real-world datasets show that LGE2LM has much better generalization capability and stronger robustness in comparison with ELM.
Fuzzy formal concept analysis (FFCA) is generally used to describe the processes of concept-cognitive learning (CCL). However, for fuzzy formal contexts, each attribute has the same weight (i.e, the ...same degree of importance) before constructing fuzzy concepts, which limits mining interesting knowledge and affects its application promotion. On the other hand, this model is hard to resist the influence of noise hidden in data, which results in poor classification learning. Moreover, the existing incremental CCL algorithms still face some challenges that the previously acquired knowledge is not fully utilized to improve the classification accuracies for dynamic data. To address these issues, we introduce different weights into fuzzy formal contexts and propose a novel incremental CCL mechanism in dynamic environment. Firstly, weight values of attributes from different decisions based on fuzzy entropy are established to measure the significant degree of attributes. Then, to comprehensively explicate the hierarchical relationships of fuzzy concepts, we construct the weighted fuzzy concept lattice and the weighted fuzzy concept space. Secondly, we design an algorithm to update the weighted fuzzy concepts for facilitating concept classification. To overcome the individual cognitive limitation, we put forward the progressive weighted fuzzy concept to remove repeated information. Furthermore, the classification prediction label and dynamic updating mechanism after adding objects are systematically discussed. Finally, we perform an experimental evaluation on ten data sets which explicate the feasibility and efficiency of our proposed approach.
Distance metric learning aims to learn a data-dependent similarity measure, which is widely employed in machine learning. Recently, metric learning algorithms that incorporate multiple kernel ...learning have shown promising outcomes for classification tasks. However, the multiple kernel learning part of the existing metric learning with multiple kernel just uses a linear combination form of different kernel functions, where each kernel shares the same weight in the entire input space, thus the potential local structure of samples located at different locations in the input space is ignored. To address the aforementioned issues, in this paper, we propose a distance metric learning approach with local multiple kernel embedding (DMLLMK) for small datasets. The weight of each kernel function in DMLLMK is assigned locally, so that there are many different values of weight in each kernel space. This local weight method enables metric learning to capture more information in the data. Our proposed DMLLMK adjusts the kernel weight by using a gating function; moreover, the kernel weight locally depends on the input data. The metric of metric learning and the parameters of the gating function are optimized simultaneously by an alternating learning process. The DMLLMK makes metric learning applicable to small datasets by constructing constraints on the set of similar pairs and dissimilar pairs such that some data are reused, and they produce different constraints on the model. In addition, regularization techniques are used to keep DMLLMK more conservative and prevent overfitting on small data. The experimental results of our proposed method when compared with other metric learning methods on the benchmark dataset show that our proposed DMLLMK is effective.
Many locations around the world have used real-time estimates of the time-varying effective reproductive number (Formula: see text) of COVID-19 to provide evidence of transmission intensity to inform ...control strategies. Estimates of Formula: see text are typically based on statistical models applied to case counts and typically suffer lags of more than a week because of the latent period and reporting delays. Noting that viral loads tend to decline over time since illness onset, analysis of the distribution of viral loads among confirmed cases can provide insights into epidemic trajectory. Here, we analyzed viral load data on confirmed cases during two local epidemics in Hong Kong, identifying a strong correlation between temporal changes in the distribution of viral loads (measured by RT-qPCR cycle threshold values) and estimates of Formula: see text based on case counts. We demonstrate that cycle threshold values could be used to improve real-time Formula: see text estimation, enabling more timely tracking of epidemic dynamics.