Continuum versus discrete networks,graph Laplacians,and reproducing kernel Hilbert spaces

Motivated by applications to machine learning, we construct a reversible and irreducible Markov chain whose state space is a certain collection of measurable sets of a chosen l.c.h. space $\mathfrak{X}$. We study the resulting network (connected undirected graph), including transience, Royden and Riesz decompositions, and kernel factorization. We describe a construction for Hilbert spaces of signed measures which comes equipped with a new notion of reproducing kernels and there is a unique solution to a regularized optimization problem involving the approximation of $L^2$ functions by functions of finite energy. The latter has applications to machine learning (for Markov random fields, for example).     

Investigating undergraduate mathematics learners’ cognitive engagement with recorded lecture videos

The use of recorded lecture videos (RLVs) in mathematics instruction continues to advance. Prior research at the post-secondary level has indicated a tendency for RLV use in mathematics to be negatively correlated with academic performance, although it is unclear whether this is because regular users are generally weaker mathematics students or because RLV use is somehow depressing student learning. Through the lens of cognitive engagement, a quasi-experimental pre- and post-test design study was conducted to investigate the latter possibility.

Cognitive engagement was operationalized using the Revised Two-Factor Study Process Questionnaire (R-SPQ-2F), which measures learning approaches on two major scales: surface and deep. In two mathematics courses at two universities, in Australia and the UK, participants were administered the questionnaire near the course start and finish. Overall findings were similar in both contexts: a reduction in live lecture attendance coupled with a dependence on RLVs was associated with an increase in surface approaches to learning.

This study has important implications for future pedagogical development and adds to the sense of urgency regarding research into best practices using RLVs in mathematics.     

Families,experiments, and nature: Learning science through project-based learning

This study brings together the research focused on science education through project-based learning (PBL). This learning project was carried out in a rural learning community and an attempt was made to adapt to the natural resources of the area by organizing educational outings, experimental activities, and encouraging the participation of families. The overall objective is to test the effectiveness of applying the PBL teaching methodology for learning science in a rural learning community. The methodology used has been qualitative, specifically, the participating research has been used and the information has been compiled in a field notebook. The results show that the didactic proposal had good results; showing that, in conclusion, science teaching today should be inclined toward more innovative educational methodologies such as PBL.     

具有学习和退化效应的单机干扰管理问题

针对工件同时具有学习和退化效应、机器具有可用性限制这一问题,建立可预见性单机干扰管理模型。在这一模型中,工件的加工时间是既与工件所排的加工位置又与工件开始加工的时间有关的函数。同时,在生产过程中由于机器发生故障或定期维修等扰动事件导致机器在某段时间内不能加工工件。目标是在同时考虑原目标函数和由扰动造成的偏离函数的情况下,构建一个新的最优时间表序列。根据干扰度量函数的不同研究了两个问题,第一个问题的目标函数是极小化总完工时间与总误工时间的加权和;第二个问题的目标函数是极小化总完工时间与总提前时间的加权和。对于所研究的问题,首先证明了最优排序具有的性质,然后建立了相应的拟多项式时间动态规划算法。     

A weakly supervised large margin domain adaptation method for isolated handwritten digit recognition

Learning handwriting categories fail to perform well when trained and tested on data from different databases. In this paper, we propose a novel large margin domain adaptation algorithm which is able to learn a transformation between training and test datasets in addition to adapting the parameters of classifier using a few or even no training labeled samples from target handwriting dataset. Additionally, we developed a framework of ensemble projection feature learning for datasets representation as a front end for our algorithm to utilize the abundant unlabeled samples in target domain. Experiments on different handwritten digit datasets adaptations demonstrate that the proposed large margin domain adaptation algorithm achieves superior classification accuracy comparing with the state of the art methods. Quantitative evaluation of the proposed algorithm shows that semi-supervised adaptation utilizing one sample per class of target domain set reduces the error rates by 64.72% comparing with a corresponding SVM classifier.     

Learning extended tree augmented naive structures

This work proposes an extended version of the well-known tree-augmented naive Bayes (TAN) classifier where the structure learning step is performed without requiring features to be connected to the class. Based on a modification of Edmonds' algorithm, our structure learning procedure explores a superset of the structures that are considered by TAN, yet achieves global optimality of the learning score function in a very efficient way (quadratic in the number of features, the same complexity as learning TANs). We enhance our procedure with a new score function that only takes into account arcs that are relevant to predict the class, as well as an optimization over the equivalent sample size during learning. These ideas may be useful for structure learning of Bayesian networks in general. A range of experiments shows that we obtain models with better prediction accuracy than naive Bayes and TAN, and comparable to the accuracy of the state-of-the-art classifier averaged one-dependence estimator (AODE). We release our implementation of ETAN so that it can be easily installed and run within Weka.     

ABLE: Attention based learning for enzyme classification

Classifying proteins into their respective enzyme class is an interesting question for researchers for a variety of reasons. The open source Protein Data Bank (PDB) contains more than 1,60,000 structures, with more being added everyday. This paper proposes an attention-based bidirectional-LSTM model (ABLE) trained on over sampled data generated by SMOTE to analyse and classify a protein into one of the six enzyme classes or a negative class using only the primary structure of the protein described as a string by the FASTA sequence as an input. We achieve the highest F1-score of 0.834 using our proposed model on a dataset of proteins from the PDB. We baseline our model against eighteen other machine learning and deep learning networks, including CNN, LSTM, Bi-LSTM, GRU, and the state-of-the-art DeepEC model. We conduct experiments with two different oversampling techniques, SMOTE and ADASYN. To corroborate the obtained results, we perform extensive experimentation and statistical testing.     

Single-Molecule Sensing of Acidic Catecholamine Metabolites Using a Programmable Nanopore

Acidic catecholamine metabolites, which could serve as diagnostic markers for many diseases, demonstrate an importance of accurate sensing. However, they share a highly similar chemical structure, which is a challenge in the design of sensing strategies. A nanopore may be engineered to sense these metabolites in a single molecule manner. To achieve this, a recently developed programmable nano-reactor for stochastic sensing (PNRSS) technique adapted with a phenylboronic acid (PBA) adaptor was applied. Three acidic catecholamine metabolites, including 3,4-dihydroxyphenylacetic acid (DOPAC), 3,4-dihydroxymandelic acid (DHMA) and 3-methoxy-4-hydroxymandetic acid (VMA) were investigated by PNRSS. Specifically, DHMA, which contains an α-hydroxycarboxylate moiety and an adjacent cis-hydroxyl groups on its benzene ring, reports two binding modes simultaneously resolvable by PNRSS. Assisted with the high resolution of PNRSS, direct regulation of these two binding modes by pH can also be observed. A custom machine learning algorithm was also developed to achieve automatic event classification.