Existing Learning Methods Research Articles

Hypergraph Learning: Methods and Practices.

Hypergraph learning is a technique for conducting learning on a hypergraph structure. In recent years, hypergraph learning has attracted increasing attention due to its flexibility and capability in modeling complex data correlation. In this paper, we first systematically review existing literature regarding hypergraph generation, including distance-based, representation-based, attribute-based, and network-based approaches. Then, we introduce the existing learning methods on a hypergraph, including transductive hypergraph learning, inductive hypergraph learning, hypergraph structure updating, and multi-modal hypergraph learning. After that, we present a tensor-based dynamic hypergraph representation and learning framework that can effectively describe high-order correlation in a hypergraph. To study the effectiveness and efficiency of hypergraph generation and learning methods, we conduct comprehensive evaluations on several typical applications, including object and action recognition, Microblog sentiment prediction, and clustering. In addition, we contribute a hypergraph learning development toolkit called THU-HyperG.

Inferring Protein Sequence-Function Relationships with Large-Scale Positive-Unlabeled Learning.

Machine learning can infer how protein sequence maps to function without requiring a detailed understanding of the underlying physical or biological mechanisms. It is challenging to apply existing supervised learning frameworks to large-scale experimental data generated by deep mutational scanning (DMS) and related methods. DMS data often contain high-dimensional and correlated sequence variables, experimental sampling error and bias, and the presence of missing data. Notably, most DMS data do not contain examples of negative sequences, making it challenging to directly estimate how sequence affects function. Here, we develop a positive-unlabeled (PU) learning framework to infer sequence-function relationships from large-scale DMS data. Our PU learning method displays excellent predictive performance across ten large-scale sequence-function datasets, representing proteins of different folds, functions, and library types. The estimated parameters pinpoint key residues that dictate protein structure and function. Finally, we apply our statistical sequence-function model to design highly stabilized enzymes.

A view-free image stitching network based on global homography

Image stitching is a traditional but challenging computer vision task, aiming to obtain a seamless panoramic image. Recently, researchers begin to study the image stitching task using deep learning. However, the existing learning methods assume a relatively fixed view during the image capturing, thus show a poor generalization ability to flexible view cases. To address the above problem, we present a cascaded view-free image stitching network based on a global homography. This novel image stitching network does not have any restriction on the view of images and it can be implemented in three stages. In particular, we first estimate a global homography between two input images from different views. And then we propose a structure stitching layer to obtain the coarse stitching result using the global homography. In the last stage, we design a content revision network to eliminate ghosting effects and refine the content of the stitching result. To enable efficient learning on various views, we also present a method to generate synthetic datasets for network training. Experimental results demonstrate that our method can achieve almost 100% elimination of artifacts in overlapping areas at the cost of acceptable slight distortions in non-overlapping areas, compared with traditional methods. In addition, the proposed method is view-free and more robust especially in a scene where feature points are difficult to detect.

SVMs multi-class loss feedback based discriminative dictionary learning for image classification

The learning model has been popular recently due to its promising results in various image classification tasks. Many existing learning methods, especially the deep learning methods, need a large amount of training data to achieve a high accuracy of classification. Conversely, only provided with a small-size dataset, some dictionary learning (DL) methods can achieve a perfect performance on a image classification task and hence still get a lot of attention. Among these DL methods, DL based feature learning methods are the mainstream for image classification in recent years, however, most of these methods have trained a classifier independently from dictionary learning. Therefore, the features extracted by the learned dictionary may not be very proper to perform classification for the classifier. Inspired by the feedback mechanism in cybernetics, this paper proposes a novel discriminative DL framework, named support vector machines (SVMs) multi-class loss feedback based discriminative dictionary learning (SMLFDL) that learns a discriminative dictionary while training SVMs to make the features extracted by the learned dictionary and SVMs better matched with each other. Because of integrating dictionary learning and SVMs training into a unified learning framework and good exactness of the looped multi-class loss term formulated from the feedback viewpoint for the classification scheme, better classification performance can be achieved. Experimental results on several widely used image databases show that SMLFDL can achieve a competitive performance with other state-of-the-art dictionary learning methods.

A Group-Based Distance Learning Method for Semisupervised Fuzzy Clustering.

Learning a proper distance for clustering from prior knowledge falls into the realm of semisupervised fuzzy clustering. Although most existing learning methods take prior knowledge (e.g., pairwise constraints) into account, they pay little attention to local knowledge of data, which, however, can be utilized to optimize the distance. In this article, we propose a novel distance learning method, which learns from the Group-level information, for semisupervised fuzzing clustering. We first present a new format of constraint information, called Group-level constraints, by elevating the pairwise constraints (must-links and cannot-links) from point level to Group level. The Groups, generated around data points contained in the pairwise constraints, carry not only the local information of data (the relation between close data points) but also more background information under some given limited prior knowledge. Then, we propose a novel method to learn a distance by using the Group-level constraints, namely, Group-based distance learning, in order to optimize the performance of fuzzy clustering. The distance learning process aims to pull must-link Groups as close as possible while pushing cannot-link Groups as far as possible. We formulate the learning process with the weights of constraints by invoking some linear and nonlinear transformations. The linear Group-based distance learning method is realized by means of semidefinite programming, and the nonlinear learning method is realized by using the neural network, which can explicitly provide nonlinear mappings. Experimental results based on both synthetic and real-world datasets show that the proposed methods yield much better performance compared to other distance learning methods using pairwise constraints.

Improved software defect prediction using Pruned Histogram-based isolation forest

Software defect prediction (SDP) is a hot topic in the modern software engineering research community. It has been used for evaluating software quality and reliability and allocating limited testing resources effectively. Based on analyzing the software source code and development process and extracting critical metrics, many data mining and machine learning methods have been used for SDP. However, these existing learning methods have difficulty with handling the imbalanced data distribution of accumulated training dataset. Isolation forest, an anomaly detection method based on the ensemble learning, has been studied to deal with the imbalanced data distribution issue for obtaining high prediction performance. However, the isolation forest method suffers from a main drawback of slow convergence, which is caused by selecting the feature value at random during the process of building isolation trees. To conquer this problem, in this paper histogram is constructed for the value set of selected isolation feature helping identify feature values preferable to build isolation trees. Motivated by the “many could be better than all” principle in the ensemble learning, the ensemble pruning strategy is further employed to optimize the obtained isolation forest, leading to a novel SDP method named PHIForest (Pruned Histogram-based Isolation Forest) in this work. The proposed method can provide fast convergence through the histogram-based splitting feature value selection, and decrease the ensemble scale and improve prediction performance through the ensemble pruning. Comprehensive experiments conducted on ten real datasets are performed to demonstrate effectiveness of the proposed SDP method.

Safe Exploration for Optimizing Contextual Bandits

Contextual bandit problems are a natural fit for many information retrieval tasks, such as learning to rank, text classification, recommendation, and so on. However, existing learning methods for contextual bandit problems have one of two drawbacks: They either do not explore the space of all possible document rankings (i.e., actions) and, thus, may miss the optimal ranking, or they present suboptimal rankings to a user and, thus, may harm the user experience. We introduce a new learning method for contextual bandit problems, Safe Exploration Algorithm (SEA), which overcomes the above drawbacks. SEA starts by using a baseline (or production) ranking system (i.e., policy), which does not harm the user experience and, thus, is safe to execute but has suboptimal performance and, thus, needs to be improved. Then SEA uses counterfactual learning to learn a new policy based on the behavior of the baseline policy. SEA also uses high-confidence off-policy evaluation to estimate the performance of the newly learned policy. Once the performance of the newly learned policy is at least as good as the performance of the baseline policy, SEA starts using the new policy to execute new actions, allowing it to actively explore favorable regions of the action space. This way, SEA never performs worse than the baseline policy and, thus, does not harm the user experience, while still exploring the action space and, thus, being able to find an optimal policy. Our experiments using text classification and document retrieval confirm the above by comparing SEA (and a boundless variant called BSEA) to online and offline learning methods for contextual bandit problems.

Supervised learning in spiking neural networks with synaptic delay-weight plasticity

Spiking neurons encode information through their spiking temporal patterns. Although the precise spike-timing based encoding scheme has long been recognised, the exact mechanism that underlies the learning of such precise spike-timing in the brain remains an open question. Most of the existing learning methods for spiking neurons are based on synaptic weight adjustment. However, biological evidences suggest that synaptic delays can also be modulated to play an important role in the learning process. This paper investigates the viability of integrating synaptic delay plasticity into supervised learning and proposes a novel learning method that adjusts both the synaptic delays and weights of the learning neurons to make them fire precisely timed spikes, that is referred to as synaptic delay-weight plasticity. Remote Supervised Method (ReSuMe) and Perceptron Based Spiking Neuron Learning Rule (PBSNLR), two representative supervised learning methods, are studied to illustrate how the synaptic delay-weight plasticity works. The performance of the proposed learning method is thoroughly evaluated on synthetic data and is further demonstrated on real-world classification tasks. The experiments show that the synaptic delay-weight learning method outperforms the traditional synaptic weight learning methods in many ways.

New Efficient Multi-Spike Learning for Fast Processing and Robust Learning

Spiking neural networks (SNNs) are considered to be more biologically plausible and lower power consuming than traditional artificial neural networks (ANNs). SNNs use discrete spikes as input and output, but how to process and learn these discrete spikes efficiently and accurately still remains a challenging task. Moreover, most existing learning methods are inefficient with complicated neuron dynamics and learning procedures being involved. In this paper, we propose efficient alternatives by firstly introducing a simplified and efficient neuron model. Based on it, we develop two new multi-spike learning rules together with an event-driven scheme being presented to improve the processing efficiency. We show that, with the as-proposed rules, a single neuron can be trained to successfully perform challenging tasks such as multi-category classification and feature extraction. Our learning methods demonstrate a significant robustness against various strong noises. Moreover, experimental results on some real-world classification tasks show that our approaches yield higher efficiency with less requirement on computation resource, highlighting the advantages and potential of spike-based processing and driving more efforts towards neuromorphic computing.

Enhanced Network Representation Learning With Community Aware and Relational Attention

Network representation learning is proposed to make it easier to perform complex inference processes on large-scale networks. It aims to represent each node in the network as a low-dimensional potential representation while preserving the structure and inherent features of the network. Most existing learning methods do not consider the comprehensive structure or the rich semantics of between nodes, which lead to incomplete embeddings. We attempt to find a way to learn network representation while keeping the network structure and relations between nodes to address this issue. In this paper, we propose a Community Aware and Relational Attention (CARA) method to enhance network representation learning. In CARA, community aware random walks capture both the community and local context from a global perspective to enhance the structure embedding. And we design a two-layer attention mechanism to model the semantic relationships between nodes, which can learn different text embedding for different neighbors. Then, these two embeddings are combined together to obtain the node representation. Experiments on four real-world networks show that CARA has better performance than other methods in link prediction and node classification.

Large-scale dynamic social data representation for structure feature learning

The problems caused by network dimension disasters and computational complexity have become an important issue to be solved in the field of social network research. The existing methods for network feature learning are mostly based on static and small-scale assumptions, and there is no modified learning for the unique attributes of social networks. Therefore, existing learning methods cannot adapt to the dynamic and large-scale of current social networks. Even super large scale and other features. This paper mainly studies the feature representation learning of large-scale dynamic social network structure. In this paper, the positive and negative damping sampling of network nodes in different classes is carried out, and the dynamic feature learning method for newly added nodes is constructed, which makes the model feasible for the extraction of structural features of large-scale social networks in the process of dynamic change. The obtained node feature representation has better dynamic robustness. By selecting the real datasets of three large-scale dynamic social networks and the experiments of dynamic link prediction in social networks, it is found that DNPS has achieved a large performance improvement over the benchmark model in terms of prediction accuracy and time efficiency. When the α value is around 0.7, the model effect is optimal.

Utilization of virtual reality content for laboratory practicum learning

At present, the learning model with practicum learning methods in the laboratory is considered as an effective learning model. However, from these advantages, there are weaknesses. One of the weaknesses of practicum learning methods is the high cost of learning because it requires much equipment. Digital Learning is an educational concept that utilizes information technology in the teaching and learning process. The concept of learning in digital multimedia is by utilizing virtual reality technology. The concept of virtual reality that simulates the real world into the digital world can be used as a solution to overcome the weakness of the existing learning method. This study provides the development of virtual reality for practicum learning in the laboratory. The knowledge base applied came from scientific experts and used as a rule in learning. The purpose of developing this application is as a prototyping framework of virtual learning that can be applied to laboratory practicum learning.

Pelatihan Aplikasi Penggunaan Matlab Untuk Meningkatkan Pengetahuan Guru

Nowadays the task of the teacher is felt to be getting tougher with the change in the curriculum used, which is to become a Competency Based Curriculum (CBC) with one of the existing learning methods namely Contextual Teaching Learning (CTL) learning. A learning using computer aids becomes an important requirement in learning with the CTL system, so it becomes a necessity for teachers to be able to master computer software that can help a learning. Software that can help learning mathematics in the classroom is MATLAB or Matrix Laboratory software, which is currently being used by teachers in mathematics and vocational engineering. To improve the ability of teachers to utilize computers in learning, especially in the use of MATLAB software, a "Training on the Use of Matlab Applications to Increase Teacher Knowledge" was held.

Learning Disentangled Representation with Pairwise Independence

Unsupervised disentangled representation learning is one of the foundational methods to learn interpretable factors in the data. Existing learning methods are based on the assumption that disentangled factors are mutually independent and incorporate this assumption with the evidence lower bound. However, our experiment reveals that factors in real-world data tend to be pairwise independent. Accordingly, we propose a new method based on a pairwise independence assumption to learn the disentangled representation. The evidence lower bound implicitly encourages mutual independence of latent codes so it is too strong for our assumption. Therefore, we introduce another lower bound in our method. Extensive experiments show that our proposed method gives competitive performances as compared with other state-of-the-art methods.

PEMANFAATAN INSTANT MESSAGING UNTUK APLIKASI PEMBELAJARAN MEMBACA AL-QUR’AN DENGAN METODE TSAQIFA

Indonesia is predominantly Muslim country. However, around 54% of Muslims cannot read the Qur'an. Many teenagers and adults argue that this is due to lack of time, shame to learn, not having a teacher and so on. Based on these problems, many methods are formulated to help the community in learning. The Tsaqifa method is one method that is claimed to make people able to read the Qur'an in just 7.5 hours. However, along with technological advances, this method began to be considered conventional. Therefore, there needs to be a new alternative so that learning methods can keep up with technological advances. Utilization of instant messaging can be used as an alternatives to implement existing learning methods especially Tsaqifa method that is expected can help communities learn independently, quickly, easily and interactively. In this study, the application of the Tsaqifa method was carried out by developing an application of chatbot in learning the Al-Qur'an reading media that utilizes the LINE Messaging API. Applications that have been built are tested with 2 methods that is Blackbox Testing and System Usability Scale (SUS). In testing using Blackbox Testing, the overall functionality of the application has been declared successfully run and based on the results of the System Usability Scale (SUS) test, obtained an average score of 73.3 out of 20 respondents which means the application can be accepted by the public (acceptable).

Machine learning based approach for vestibular disorder diagnostic in videonystagmography

Nystagmus is a common sign of peripheral vestibular disorder (VD). Several problems have been recently noted from Video Nystagmo Graphic (VNG) analysis to get relevant diagnosis of VD diseases. The Vestibulo-Ocular Response (VOR) is characterized by a smooth pursuit eye movement in one direction, called slow phase of ocular nystagmus, which is interrupted by saccades (fast phases) in the other direction. The recording of ocular nystagmus during vestibular tests does not quantify the true response of the vestibulo-ocular reflex (VOR). So, to extract the real VOR, our study is focused on nystagmus analysis with videonystagmography (VNG) technique based on measuring amplitude and frequency vibration of eyeball movement. In this paper, we have proposed a fully automatic system based on nystagmus parameter analysis using a pupil detection algorithm and a machine learning approach for VD recognition. Firstly, an estimation of the pupil movement vectors using Hough Transform (HT) is employed to approximate the location of pupil region. Then, temporal and frequency features are computed from the rotation angle variation of the pupil motion. Finally, pertinent features are selected using a statistical criterion for discrimination and classification of the VD disease. Experimental results are employed using two categories which are normal and pathological cases. By discriminating the reduced features with the Support Vector Machine (SVM) technique, 94% of classification accuracy results are achieved. Compared to existing learning methods, the proposed technique is extremely effective to resolve the problem of VD assessment and provide an accurate diagnosis for medical devices.

Online Identification of Nonlinear Stochastic Spatiotemporal System With Multiplicative Noise by Robust Optimal Control-Based Kernel Learning Method.

In this paper, we propose a novel kernel method for the online identification of stochastic nonlinear spatiotemporal dynamical systems using the robust control approach. By the difference method, the stochastic spatiotemporal (SST) systems driven by multiplicative noise are first transformed into a class of multi-input-multi-output-partially linear kernel models (PLKMs) with heterogeneous random terms. With the help of techniques established for reproducing kernel Hilbert space, the online learning problem is reasonably considered as an output feedback control problem for a group of time varying linear dynamical systems. We develop an effective algorithm to address the learning problem of PLKM and SST systems by employing the model predictive control theory. Compared with the existing learning methods, the new one can achieve adaptive, robust, and fast convergent online modeling performance for the spatiotemporal dynamics with multiplicative noise, which greatly facilitates the characterization of physical characteristics of the system. Moreover, this investigation for the first time addresses the learning problems for SST systems with novel robust control techniques, which can provide some novel insights into the design of kernel machine learning methods from the perspective of optimal control theory. Numerical studies for benchmark systems are presented to illustrate the effectiveness and efficiency of our new method.

A Highly Effective and Robust Membrane Potential-Driven Supervised Learning Method for Spiking Neurons.

Spiking neurons are becoming increasingly popular owing to their biological plausibility and promising computational properties. Unlike traditional rate-based neural models, spiking neurons encode information in the temporal patterns of the transmitted spike trains, which makes them more suitable for processing spatiotemporal information. One of the fundamental computations of spiking neurons is to transform streams of input spike trains into precisely timed firing activity. However, the existing learning methods, used to realize such computation, often result in relatively low accuracy performance and poor robustness to noise. In order to address these limitations, we propose a novel highly effective and robust membrane potential-driven supervised learning (MemPo-Learn) method, which enables the trained neurons to generate desired spike trains with higher precision, higher efficiency, and better noise robustness than the current state-of-the-art spiking neuron learning methods. While the traditional spike-driven learning methods use an error function based on the difference between the actual and desired output spike trains, the proposed MemPo-Learn method employs an error function based on the difference between the output neuron membrane potential and its firing threshold. The efficiency of the proposed learning method is further improved through the introduction of an adaptive strategy, called skip scan training strategy, that selectively identifies the time steps when to apply weight adjustment. The proposed strategy enables the MemPo-Learn method to effectively and efficiently learn the desired output spike train even when much smaller time steps are used. In addition, the learning rule of MemPo-Learn is improved further to help mitigate the impact of the input noise on the timing accuracy and reliability of the neuron firing dynamics. The proposed learning method is thoroughly evaluated on synthetic data and is further demonstrated on real-world classification tasks. Experimental results show that the proposed method can achieve high learning accuracy with a significant improvement in learning time and better robustness to different types of noise.

Efficient and Robust Supervised Learning Algorithm for Spiking Neural Networks

Recent advances in neuroscience have revealed that the brains neural information is encoded via precisely timed spike trains, but not just because of the neural firing rate. Owing to the inherent complexity of processing spike sequences, the formulation of an efficiently supervised learning algorithm is difficult and remains an important problem in research. This paper proposes an efficient and robust membrane potential-driven (ERMPD) supervised learning method capable of training neurons to generate desired sequences of spikes. For efficiency, at undesired output times, ERMPD calculates the membrane potential and makes a comparison to a firing threshold only at specified time points. For robustness, the dynamic threshold can be applied by the existing supervised learning methods to improve noise tolerance. Experimental results show that the proposed method has higher learning accuracy and efficiency over existing learning methods. Thus, it is more powerful for solving complex and real-time problems.

METODE PEMBELAJARAN BAHASA ARAB DI PONDOK PESANTREN TRADISIONAL

Arabic language is must to learn in pesantren, because Arabic language used in a classic books in Arabic, the actifity of worship is done daily also with the Arabic language. Pesantren is the only educational institution studying Arabic in the old of the development of Islam, even today. Arabic learning methods used by pesantren, especially traditional pesantren have been decades old using qawaid wa tarjamah method. The method does this even become a grip or characteristic of the learning process undertaken in traditional pesantren, since the majority assume that the existing learning method is the best method in the learning process, thus overriding new methods or the existence of other methods which is likely to be more effectively used in the learning of the Arabic language. There needs to be a review of the learning methods currently used by boarding schools, so that learning Arabic language can facilitate the learner or santri.