Performance In Learning Tasks Research Articles

Blind color deconvolution, normalization, and classification of histological images using general super Gaussian priors and Bayesian inference

Background and Objective:Color variations in digital histopathology severely impact the performance of computer-aided diagnosis systems. They are due to differences in the staining process and acquisition system, among other reasons. Blind color deconvolution techniques separate multi-stained images into single stained bands which, once normalized, can be used to eliminate these negative color variations and improve the performance of machine learning tasks. Methods:In this work, we decompose the observed RGB image in its hematoxylin and eosin components. We apply Bayesian modeling and inference based on the use of Super Gaussian sparse priors for each stain together with prior closeness to a given reference color-vector matrix. The hematoxylin and eosin components are then used for image normalization and classification of histological images. The proposed framework is tested on stain separation, image normalization, and cancer classification problems. The results are measured using the peak signal to noise ratio, normalized median intensity and the area under ROC curve on five different databases. Results:The obtained results show the superiority of our approach to current state-of-the-art blind color deconvolution techniques. In particular, the fidelity to the tissue improves 1,27 dB in mean PSNR. The normalized median intensity shows a good normalization quality of the proposed approach on the tested datasets. Finally, in cancer classification experiments the area under the ROC curve improves from 0.9491 to 0.9656 and from 0.9279 to 0.9541 on Camelyon-16 and Camelyon-17, respectively, when the original and processed images are used. Furthermore, these figures of merits are better than those obtained by the methods compared with. Conclusions:The proposed framework for blind color deconvolution, normalization and classification of images guarantees fidelity to the tissue structure and can be used both for normalization and classification. In addition, color deconvolution enables the use of the optical density space for classification, which improves the classification performance.

Debiasing Causal Inferences: Over and Beyond Suboptimal Sampling

Background We have previously presented two educational interventions aimed to diminish causal illusions and promote critical thinking. In both cases, these interventions reduced causal illusions developed in response to active contingency learning tasks, in which participants were able to decide whether to introduce the potential cause in each of the learning trials. The reduction of causal judgments appeared to be influenced by differences in the frequency with which the participants decided to apply the potential cause, hence indicating that the intervention affected their information sampling strategies. Objective In the present study, we investigated whether one of these interventions also reduces causal illusions when covariation information is acquired passively. Method Forty-one psychology undergraduates received our debiasing intervention, while 31 students were assigned to a control condition. All participants completed a passive contingency learning task. Results We found weaker causal illusions in students that participated in the debiasing intervention, compared to the control group. Conclusion The intervention affects not only the way the participants look for new evidence, but also the way they interpret given information. Teaching implications Our data extending previous results regarding evidence-based educational interventions aimed to promote critical thinking to situations in which we act as mere observers.

Calmodulin acetylation: A modification to remember

The formation of new memories appears to require alterations in the shape and strength of synapses within the hippocampus, yet our knowledge of the molecular mechanisms underlying these changes remains incomplete. Zhang and colleagues provide new understanding of memory formation by uncovering the lysine acetyltransferase SRC3 as the key driver of the novel posttranslational modification of calmodulin (CaM) acetylation, which regulates CaM's activity and subsequent activation of CaMKII. This new pathway is demonstrated to be both necessary and sufficient for CA3→CA1 synapse long-term potentiation (LTP) and fear memory formation, and this approach may act as a blueprint for future investigation of the role of acetylation of other proteins in neuronal functions.

Joint nonlinear feature selection and continuous values regression network

The curse of dimensionality is a long-standing intractable issue in many machine learning and computer vision tasks. Feature selection, a data preprocessing technique, aims to select most discriminative feature subsets for improving performance of downstream machine learning tasks. However, most existing feature selection methods concentrate upon learning a linear relationship between data points and their labels, which causes that they are incapable of handling nonlinear complex data in real-world applications. In this paper, we first propose a novel end-to-end Nonlinear Feature Selective Networks (NFSN) that be able to select discriminative feature subsets while preserving their nonlinear structure by embedding a ℓ2,p-norm regularized hidden layer into designed continuous values regression networks. In addition, we propose an efficient optimization algorithm that joins back propagation algorithm and re-weighted optimization strategy to acquire derivative of all weights accurately. Experimental results on the nonlinear analog pulse signal and real-world datasets demonstrate the superiority of proposed method compared to some related methods on feature selection. Our source code available on: https://github.com/StevenWangNPU/NFSN.

The study of virtual reality adaptive learning method based on learning style model

AbstractLearning style is the endogenous cause of students' unique behaviors when they are performing learning tasks. The adaptive learning system that considers learning style can provide a personalized experience to stimulate students' enthusiasm, which had been widely studied in recent years. However, most of such existing systems are constructed based on a desktop environment, which leads to the less‐than‐ideal effect of personalized learning due to the limitation of interaction means and environmental dimension. Therefore, an adaptive virtual reality learning method based on the learning style model was proposed in this study. This method continuously iterated the identification of learning style based on students' subjective and objective data. Then, the content of virtual learning environment was sustainedly adjusted according to the identification results, thus enabling the environment to dynamically adapt to students' learning styles. To evaluate the feasibility of this proposed method, a controlled experiment on 152 participants was conducted. Results show that this method obtained relatively stable and accurate results of learning style identification, with an accuracy range of 74.38%–80.30%. Moreover, learning in the virtual environment constructed based on this method had a positive impact on students' learning motivation and outcomes.

Knowledge Preserving and Distribution Alignment for Heterogeneous Domain Adaptation

Domain adaptation aims at improving the performance of learning tasks in a target domain by leveraging the knowledge extracted from a source domain. To this end, one can perform knowledge transfer between these two domains. However, this problem becomes extremely challenging when the data of these two domains are characterized by different types of features, i.e., the feature spaces of the source and target domains are different, which is referred to as heterogeneous domain adaptation (HDA). To solve this problem, we propose a novel model called Knowledge Preserving and Distribution Alignment (KPDA), which learns an augmented target space by jointly minimizing information loss and maximizing domain distribution alignment. Specifically, we seek to discover a latent space, where the knowledge is preserved by exploiting the Laplacian graph terms and reconstruction regularizations. Moreover, we adopt the Maximum Mean Discrepancy to align the distributions of the source and target domains in the latent space. Mathematically, KPDA is formulated as a minimization problem with orthogonal constraints, which involves two projection variables. Then, we develop an algorithm based on the Gauss–Seidel iteration scheme and split the problem into two subproblems, which are solved by searching algorithms based on the Barzilai–Borwein (BB) stepsize. Promising results demonstrate the effectiveness of the proposed method.

Attention-based Sequence to Sequence Learning for Machine Translation of Low Resourced Indic Languages – A case of Sanskrit to Hindi

Deep Learning techniques are powerful in mimicking humans in a particular set of problems. They have achieved a remarkable performance in complex learning tasks. Deep learning inspired Neural Machine Translation (NMT) is a proficient technique that outperforms traditional machine translation. Performing machine-aided translation on Indic languages has always been a challenging task considering their rich and diverse grammar. The neural machine translation has shown quality results compared to the traditional machine translation approaches. The fully automatic machine translation becomes problematic when it comes to low-resourced languages, especially with Sanskrit. This paper presents attention mechanism based neural machine translation by selectively focusing on a particular part of language sentences during translation. The work shows the construction of Sanskrit to Hindi bilingual parallel corpus with nearly 10K samples and having 178,000 tokens. The neural translation model equipped with an attention mechanism has been trained on Sanskrit to Hindi parallel corpus. The approach has shown the significance of attention mechanisms to overcome long-term dependencies, primarily associated with low resources Indic languages. The paper shows the attention plots on testing data to demonstrate the alignment between source and translated words. For the evaluation of the translated sentences, manual score based human evaluation and automatic evaluation metric based techniques have been adopted. The attention mechanism based neural translation has achieved 88% accuracy in human evaluation and a BLEU score of 0.92 on Sanskrit to Hindi translation.

Few-Shot Learning in Spiking Neural Networks by Multi-Timescale Optimization.

Learning new concepts rapidly from a few examples is an open issue in spike-based machine learning. This few-shot learning imposes substantial challenges to the current learning methodologies of spiking neuron networks (SNNs) due to the lack of task-related priori knowledge. The recent learning-to-learn (L2L) approach allows SNNs to acquire priori knowledge through example-level learning and task-level optimization. However, existing L2L-based frameworks do not target the neural dynamics (i.e., neuronal and synaptic parameter changes) on different timescales. This diversity of temporal dynamics is an important attribute in spike-based learning, which facilitates the networks to rapidly acquire knowledge from very few examples and gradually integrate this knowledge. In this work, we consider the neural dynamics on various timescales and provide a multi-timescale optimization (MTSO) framework for SNNs. This framework introduces an adaptive-gated LSTM to accommodate two different timescales of neural dynamics: short-term learning and long-term evolution. Short-term learning is a fast knowledge acquisition process achieved by a novel surrogate gradient online learning (SGOL) algorithm, where the LSTM guides gradient updating of SNN on a short timescale through an adaptive learning rate and weight decay gating. The long-term evolution aims to slowly integrate acquired knowledge and form a priori, which can be achieved by optimizing the LSTM guidance process to tune SNN parameters on a long timescale. Experimental results demonstrate that the collaborative optimization of multi-timescale neural dynamics can make SNNs achieve promising performance for the few-shot learning tasks.

Repetitive transcranial magnetic stimulation during a spatial memory task leads to a decrease in brain metabolic activity

Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive neuromodulation technique that is able to generate causal-based interferences between brain networks and cognitive or behavioral responses. It has been used to improve cognition in several disease models. However, although its exploration in healthy animals is essential to attribute its pure effect in learning and memory processes, studies in this regard are scarce. We aimed to evaluate whether rTMS leads to memory facilitation in healthy rats, and to explore the brain-related oxidative metabolism. We stimulated healthy Wistar rats with a high-frequency (100 Hz) and low-intensity (0.33 T) protocol during three consecutive days and evaluated the effect on the performance of an allocentric spatial reference learning and memory task. Following the last day of learning, we assessed oxidative brain metabolism through quantitative cytochrome c oxidase (CCO) histochemistry. The results showed that rTMS did not improve spatial memory in healthy rats, but the behavioral outcome was accompanied by a CCO reduction in the prefrontal, retrosplenial, parietal, and rhinal cortices, as well as in the striatum, amygdala, septum, mammillary bodies, and the hippocampus, reflecting a lower metabolic activity. In conclusion, rTMS induces a highly efficient use of brain regions associated with spatial memory.

Parent-Teacher-Learner Collaboration in Modular Distance Learning

In the advent of pandemic due to COVID 19, modular distance learning became the mostly implemented approach in many schools in the Philippines – the school prepares and distributes learning materials, students stay at home and perform learning tasks, while parents assist in supervision. This study dealt with the role of parents, teachers and learners as they collaborate and bring about the best in education amidst pandemic. Fifteen respondents constituting of parents, teachers and learners participated in this qualitative research. Ten questions were solicited through Google forms on a weekly interval as the modular distance learning was being implemented in Talangan Integrated National High School, Nagcarlan, Laguna, Philippines. Qualitative questions pertaining to parents’ assistance and supervision, preparation of module and supplemental materials, provision of feedbacks, and learning habits were sent for the respondents to share ideas based on their experiences. Data were encoded, processed through word cloud generator, interpreted, and analyzed thematically in order to construct conclusion about meaningful collaboration amidst pandemic. Themes developed encompassed parents, teachers and learners contribution in collaboration. Parents’ hindrances to collaboration can be classified as internal and external in which, both can be mitigated through school intervention. Teachers can motivate parents in the process of collaboration when they reach out, give quality time and show real concern for learners’ interest. Learners are advised to avoid cramming, procrastinating, distractions, and set goals for themselves associated with every modular lesson. Recommendations were laid down to school heads and community partners, institutionalizing modular distance learning if pandemic still prevails.

A Unified Scheme for Distance Metric Learning and Clustering via Rank-Reduced Regression

Distance metric learning aims to learn a positive semidefinite matrix such that similar samples are preserved with small distances while dissimilar ones are mapped with big values above a predefined margin. It can facilitate to improve the performance of certain learning tasks. In this article, distance metric learning and clustering are integrated into an unified framework via rank-reduced regression. First, distance metric learning is proved to be consistent with rank-reduced regression, which provides a new perspective to learn structured regularization matrices. Second, orthogonal and non-negative rank-reduced regression problems are addressed individually for clustering, and the corresponding algorithms with proved convergence are proposed. Finally, both distance metric learning and clustering are addressed simultaneously in the problem formulation, which may trigger some new insights for learning an effective clustering oriented low-dimensional embedding. To show the superior performance of the proposed method, we compare it with several state-of-the-art clustering approaches. And, extensive experiments on the test datasets demonstrate the superiority of the proposed method.

Multi-Label Classification of Research Papers Using Multi-Label K-Nearest Neighbour Algorithm

With the frequent interaction and cooperation between different disciplines in recent years, the number of research papers associated with multiple subjects increased. Correspondingly, some of the existing literatures belong to a single discipline, while others may simultaneously involve more than 2 subjects. At this time, the traditional single-label text classification is not conducive to people obtaining comprehensive and cutting-edge research papers in real life. Thus, it’s of great importance to conduct a multi-label classification of research papers effectively. This paper tests the performance of multi-label learning tasks with text data obtained from the Kaggle website. Firstly, lemmatization and Term Frequency-Inverse Document Frequency (TF-IDF) are used for feature extraction in the pre-processing part. The critical information of text content is statistically analysed, and text content is converted into numerical and high-dimensional vector space. As the traditional single-label classification algorithm is not suitable for the above problem, this paper adopts the Multi-Label K-Nearest Neighbour (ML-KNN) algorithm framework for classification. Experimental results report that the ML-KNN algorithm has achieved better results in multi-label text classification problems than a traditional multi-label algorithm, which proves the effectiveness of the ML-KNN algorithm for text data prediction with multiple subjects. Moreover, the work in this paper is analysed and summarized.

Learning Representation From Concurrence-Words Graph For Aspect Sentiment Classification

Abstract Aspect sentiment classification is an important research topic in natural language processing and computational linguistics, assisting in automatically review analysis and emotional tendency judgement. Different from extant methods that focus on text sequence representations, this paper presents a network framework to learn representation from concurrence-words relation graph (LRCWG), so as to improve the Macro-F1 and accuracy. The LRCWG first employs the multi-head attention mechanism to capture the sentiment representation from the sentences which can learn the importance of text sequence representation. And then, it leverages the priori sentiment dictionary information to construct the concurrence relations of sentiment words with Graph Convolution Network (GCN). This assists in that the learnt context representation can keep both the semantics integrity and the features of sentiment concurrence-words relations. The designed algorithm is experimentally evaluated with all the five benchmark datasets and demonstrated that the proposed aspect sentiment classification can significantly improve the prediction performance of learning task.

Oxidative Stress and Cognitive Alterations Induced by Cancer Chemotherapy Drugs: A Scoping Review.

Cognitive impairment is one of the most deleterious effects of chemotherapy treatment in cancer patients, and this problem sometimes remains even after chemotherapy ends. Common classes of chemotherapy-based regimens such as anthracyclines, taxanes, and platinum derivatives can induce both oxidative stress in the blood and in the brain, and these effects can be reproduced in neuronal and glia cell cultures. In rodent models, both the acute and repeated administration of doxorubicin or adriamycin (anthracyclines) or cisplatin impairs cognitive functions, as shown by their diminished performance in different learning and memory behavioural tasks. Administration of compounds with strong antioxidant effects such as N-acetylcysteine, gamma-glutamyl cysteine ethyl ester, polydatin, caffeic acid phenethyl ester, and 2-mercaptoethane sulfonate sodium (MESNA) counteract both oxidative stress and cognitive alterations induced by chemotherapeutic drugs. These antioxidant molecules provide the scientific basis to design clinical trials in patients with the aim of reducing the oxidative stress and cognitive alterations, among other probable central nervous system changes, elicited by chemotherapy in cancer patients. In particular, N-acetylcysteine and MESNA are currently used in clinical settings and are therefore attracting scientific attention.

Efficient-PrototypicalNet with self knowledge distillation for few-shot learning

The focus of recent few-shot learning research has been on the development of learning methods that can quickly adapt to unseen tasks with small amounts of data and low computational cost. In order to achieve higher performance in few-shot learning tasks, the generalizability of the method is essential to enable it generalize well from seen tasks to unseen tasks with limited number of samples. In this work, we investigate a new metric-based few-shot learning framework which transfers the knowledge from another effective classification model to produce well generalized embedding and improve the effectiveness in handling unseen tasks. The idea of our proposed Efficient-PrototypicalNet involves transfer learning, knowledge distillation, and few-shot learning. We employed a pre-trained model as a feature extractor to obtain useful features from tasks and decrease the task complexity. These features reduce the training difficulty in few-shot learning and increase the performance. Besides that, we further apply knowledge distillation to our framework and achieve extra performance improvement. The proposed Efficient-PrototypicalNet was evaluated on five benchmark datasets, i.e., Omniglot, miniImageNet, tieredImageNet, CIFAR-FS, and FC100. The proposed Efficient-PrototypicalNet achieved the state-of-the-art performance on most datasets in the 5-way K-shot image classification task, especially on the miniImageNet dataset.

Adaptive graph-based generalized regression model for unsupervised feature selection

Unsupervised feature selection is an important method to reduce dimensions of high-dimensional data without labels, which is beneficial to avoid “curse of dimensionality” and improve the performance of subsequent machine learning tasks, like clustering and retrieval. A lot of methods have been developed in recent years. However, most of them only focus on selecting discriminative features but ignore the redundancy of features. Besides, these methods construct the similarity structure of data in the original input space and remain fixed during feature selection. The similarity structure is easily affected by the noises and outliers. Then the reliability of the similarity structure will be reduced in the procedure of feature selection. In order to address these issues, we present a novel generalized regression model equipped with an uncorrelated constraint and the ℓ2,1-norm regularization. It can simultaneously select the uncorrelated and discriminative features as well as reduce the variance of these data points belonging to the same neighborhood, which is helpful for the clustering task. Furthermore, to obtain the similarity structure of data and alleviate the influence of noises and outliers, we propose to adaptively learn a similarity-induced graph matrix in the reduced space and integrate it into the generalized regression model. An alternative iterative optimization algorithm is developed to solve the final objective function. A series of experiments are carried out on nine real-world data sets to demonstrate the effectiveness of the proposed method in comparison with other competing approaches.

Survey on the Application of Deep Learning in Extreme Weather Prediction

Because of the uncertainty of weather and the complexity of atmospheric movement, extreme weather has always been an important and difficult meteorological problem. Extreme weather events can be called high-impact weather, the ‘extreme’ here means that the probability of occurrence is very small. Deep learning can automatically learn and train from a large number of sample data to obtain excellent feature expression, which effectively improves the performance of various machine learning tasks and is widely used in computer vision, natural language processing, and other fields. Based on the introduction of deep learning, this article makes a preliminary summary of the existing extreme weather prediction methods. These include the ability to use recurrent neural networks to predict weather phenomena and convolutional neural networks to predict the weather. They can automatically extract image features of extreme weather phenomena and predict the possibility of extreme weather somewhere by using a deep learning framework.

Resource-constrained FPGA/DNN co-design

Deep neural networks (DNNs) have demonstrated super performance in most learning tasks. However, a DNN typically contains a large number of parameters and operations, requiring a high-end processing platform for high-speed execution. To address this challenge, hardware-and-software co-design strategies, which involve joint DNN optimization and hardware implementation, can be applied. These strategies reduce the parameters and operations of the DNN, and fit it into a low-resource processing platform. In this paper, a DNN model is used for the analysis of the data captured using an electrochemical method to determine the concentration of a neurotransmitter and the recoding electrode. Next, a DNN miniaturization algorithm is introduced, involving combined pruning and compression, to reduce the DNN resource utilization. Here, the DNN is transformed to have sparse parameters by pruning a percentage of its weights. The Lempel–Ziv–Welch algorithm is then applied to compress the sparse DNN. Next, a DNN overlay is developed, combining the decompression of the DNN parameters and DNN inference, to allow the execution of the DNN on a FPGA on the PYNQ-Z2 board. This approach helps avoid the need for inclusion of a complex quantization algorithm. It compresses the DNN by a factor of 6.18, leading to about 50% reduction in the resource utilization on the FPGA.

No Need to Reinvent ‐ Augmenting a Flipped Class Histology Curriculum with a Commercial Online Course

Histology, a pillar of anatomical sciences for both health professional and graduate students, requires sufficient time and practice to develop the necessary visual literacy and skills in synthesis to achieve competency. However, histology contact hours in medical and graduate education have reduced significantly in the past decades due to disseminated curricular changes emphasizing integration of basic sciences, clinical content and self-directed learning. In turn, adjunct online resources have been in steady incline and in the past year, in response to COVID-19, quickly became an essential component of anatomical sciences education following campus closures. Self-directed learning is challenging in normal circumstances as success is impacted by the level of student preparedness and engagement, which in turn can be dependent on the quality of the asynchronous learning resources. Although many online histology resources are available, few are comprehensive and provide step-by-step instruction in identification of structures in tissue slides. In this pilot project, the efficacy of using a commercial online histology (COH) course (DaVinci Academy, Cleveland, OH) as a primary preparatory resource in a graduate histology curriculum implementing a flipped-classroom pedagogy was assessed. During fall 2020 (F2020), first-year graduate students enrolled in a hybrid histology course were required to review learning objectives, read textbook sections, and complete the COH lessons, consisting of complete lectures, lab videos, and quizzes as a set of pre-work to prepare for on-campus and virtual classes. The synchronous class sessions had minimal didactic lectures, but instead had a series of small group active learning tasks. In a COMIRB exempt retrospective study (#20-2891), the assessment outcomes of the F2020 cohort (n=26) were compared to a fall 2019 (F2019) cohort of students (n=25) who completed the identical histology curriculum delivered in a flipped-class style by the same instructor, but without the commercial online histology course. All course quiz and exam scores were compared across the F2019 and F2020 cohorts using a Kruskal Wallis test for non-parametric data. There were no significant differences in assessment outcomes between F2019 and F2020 cohorts (p>0.05). Survey of the F2020 cohort throughout the course revealed consistently high rating of the COH with 80% of students believing the resource improved their ability to read and interpret histology slides. Interestingly, students perceived the lecture and lab videos in the COH delivered by the same in-class instructor to be more valuable than those delivered by a third-party instructor, although the assessment performance was independent of the online instructor (p>0.05). The results suggest that the online histology course can be an effective asynchronous learning resource for a histology course, allowing for more efficient use of synchronous class time to support a higher taxonomy of learning.

Embedding online activities during lecture time: Roll call or enhancement of student participation?

Student attendance has long polarized the higher education sector with reports of no to little effect on student success to positive relationships between attendance frequency at face to face and synchronous online lectures and better student engagement and achievement. This study investigates the impact of embedded online activities during lecture time on student learning by utilizing students’ portable devices to divert undesirable study behaviors such as gaming and social media activity during class. The aim of the learning intervention is to improve attendance at undergraduate engineering lectures as well as providing better connection to the subject content. Study participants were third year Bachelor of Engineering students enrolled in a mandatory “Digital System Design” course as part of their degree at a major research university in New Zealand. To explore the student experience of embedded active learning tasks on engagement and academic achievement, both qualitative and quantitative data were collected from N = 75 students over a three-year period when the course underwent a re-design utilizing a participatory action research approach. Student focus group discussions and learning analytics data revealed that the completion of online activities during lectures can lead to cognitive overload negatively affecting engagement. However, real-time feedback on learning via synchronization of learning tasks with the lecture content improved student–student and student–teacher connections and thereby contributed to a more positive overall learning experience. The role of stimulating learner motivation and attendance is discussed against Keller’s ARCS model and recommendations for teaching practice are given.