Self-learning Algorithm Research Articles

Grid-clustered rough set model for self-learning and fast reduction

Rough set theory has been playing a significant role in data mining, but deficiencies still exist in fast attribute reduction and self-learning for decision systems. Grid-clustered rough set (GCRS) model computing the universe partition based on grid subspace cluster (GSC) algorithm is constructed in this paper to fill the aforementioned gaps. GSC algorithm characterized by densities and distances in grid subspace is firstly presented along with the automatically selecting of cluster centers. Then a rough self-learning theory including extensional learning and intensional learning is raised. To realize quick attribute reduction and self-learning, a novel rough set model named GCRS that integrates rough set theory and GSC algorithm is proposed, and a corresponding quick attribute reduction algorithm and a rough self-learning algorithm based on GCRS are subsequently designed. A multitude of experiments demonstrate three aspects: a) results of attribute reduction using GCRS model are proved to be of higher classification accuracies; b) attribute reduction based on GCRS model for big data manifests faster speed; c) results of rough self-learning experiments illustrate the validity of the proposed theory.

Modeling and Self-Learning Soft-Grasp Control for Free-Floating Space Manipulator During Target Capturing Using Variable Stiffness Method

During target capturing operation, the changes in the dynamics parameters of a free-floating space manipulator degrade the performance of the base attitude stabilization. This paper presents a new self-learning soft-grasp control algorithm based on the variable stiffness technology. First, the dynamic model of variable stiffness joint space manipulator system is established. Simultaneously, the detailed dynamic analysis of pre-impact and post-impact stages is carried out. Second, a new soft-grasp control strategy utilizing cellular differential evolution algorithm combined opposition-based learning with orthogonal crossover is employed to minimize the base angular momentum. Its principle is to solve the optimal stiffness value of the variable stiffness joint to realize desired buffering. Thereafter, we put forward an adaptive backstepping sliding mode control method to track the actual joint stiffness. Finally, the proposed method is applied to a two-degree of freedom planar free-floating space manipulator and the simulation results verify the effectiveness.

Mitigation of DDoS threat to service attainability in cloud premises

Cloud environment provides users with amicable services that help them to attain their organisational and personal goals. The end user suffers from unavailability and delay of promised services under few circumstances. This is due to distributed denial of service (DDoS) attack where the illegitimate users on cloud who always generate different kinds of bots and zombies to disrupt or breakdown service availability. A DDoS detection algorithm (DDA) is proposed to trace out the existing attack by employing action-based monitoring (AM) approach on the incoming traffic. The human discovery algorithm (HDA) is implemented by performing a Turing test targeting on human to segregate the bots from the normal user. The supervisor self-learning algorithm (SSLA) is another mitigation test that allows future bots to be isolated. This model has shown improvement in terms of client success ratio, load balancing, and detection, prevention and mitigation of DDoS attack with respect to time.

Self-Learning Control Using Dual Heuristic Programming with Global Laplacian Eigenmaps

In this paper, to solve nonlinear optimal control problems which can be modeled as Markov decision processes (MDPs), we present an online self-learning control algorithm called dual heuristic programming with global Laplacian Eigenmaps (GLEM-DHP). The GLEM-DHP algorithm makes use of GLEM, which is an improved manifold learning approach with global information, to learn the features for value function approximation of the MDP. Different from traditional feature representation methods using neural networks, the manifold-based features can be learned before the online learning process by collecting samples from the MDP. More importantly, in addition to local features, global information can also be utilized by using the geodesic minimum spanning tree (GMST) approach. Based on the theoretical property of GMST, it is shown that the GLEM-based features can represent the intrinsic geometric property of MDP states, which is beneficial to improve the performance of value function approximation and, hence, leads to better learning control properties. To compare the proposed method with previous learning control algorithms, the performance of GLEM-DHP is evaluated on two nonlinear control problems, which include the cart-pole problem and the ball-plate control problem. Simulation and experimental results show that the GLEM-DHP algorithm can obtain better learning control performance than previous learning control algorithms with manually designed features, as well as manifold features only with local information.

Dynamic Cluster Formation Game for Attributed Graph Clustering.

Besides the topological structure, there are additional information, i.e., node attributes, on top of the plain graphs. Usually, these systems can be well modeled by attributed graphs, where nodes represent component actors, a set of attributes describe users' portraits and edges indicate their connections. An elusive question associated with attributed graphs is to study how clusters with common internal properties form and evolve in real-world networked systems with great individual diversity, which leads to the so-called problem of attributed graph clustering (AGC). In this paper, we comprehended AGC naturally as a dynamic cluster formation game (DCFG), where each node's feasible action set can be constrained by every cluster in a discrete-time dynamical system. Specifically, we carried out a deep research on a special case of finite dynamic games, named dynamic social game (DSG), the convergence of the finite Nash equilibrium sequence in a DSG was also proved strictly. By carefully defining the feasible action set and the utility function associated with each node, the proposed DCFG can be well related to a DSG; and we showed that a balanced solution of AGC could be found by solving a finite set of coupled static Nash equilibrium problems in the related DCFG. We, finally, proposed a self-learning algorithm, which can start from any arbitrary initial cluster configuration, and, finally, find the corresponding balanced solution of AGC, where all nodes and clusters are satisfied with the final cluster configuration. Extensive experiments were applied on real-world social networks to demonstrate both effectiveness and scalability of the proposed approach by comparing with the state-of-the-art graph clustering methods in the literature.

Empirical studies of Gaussian process based Bayesian optimization using evolutionary computation for materials informatics

Using evolutionary computation, we empirically investigate convergence properties of Gaussian process based Bayesian optimization (BO). We use evolutionary computation for the learning of the prediction model and optimization of the acquisition function (auxiliary search) of BO. For practical use for materials informatics, we address three issues in BO: (1) the stopping conditions, (2) the initial data size, and (3) the unknown smoothness of the target function. Then, we introduce a goal-directed acquisition function in which a target value as a desired property of a compound is incorporated. In addition, we present an ensemble method of BO, in which each BO in the ensemble has a random property and a kernel function with a different smoothness. Experimental results for the materials data sets on melting points of binary compounds and hydrogen weight percentages of hydrogen storage materials with two to four constituent elements show the effectiveness of the ensemble method of BO. Additionally, using an ensemble of BOs presents that the obtained results (increase of the number of samples acquired) are not simply a result of additional BOs. The goal-directed acquisition function and the ensemble of BOs which we propose should be techniques that can be used in the realization of a new materials recommendation system with a self-learning algorithm. Due to the self-learning algorithm realized by BO, the property prediction performance of the algorithm would be increasingly improved.

Improving the evaluation of soundscape variability via blind source separation

Evaluation of soundscape variability is essential for acoustic-based biodiversity monitoring. To study biodiversity change, many researchers tried to quantify the complexity of biological sound. However, the analysis of biological sound remains difficult because the soundscape is made up of multiple sound sources. To facilitate the acoustic analysis, we have applied non-negative matrix factorization (NMF) to separate different sound sources in an unsupervised manner. NMF is a self-learning algorithm which factorizes a non-negative matrix as a basis matrix and an encoding matrix. Based on the periodicity information learned from the encoding matrix, biological chorus and the other noise sources can be efficiently separated. Besides, vocalizations of different species can also be separated by using the encoding information learned from multiple layers of NMF and convolutive NMF. In this presentation, we will demonstrate the application of NMF-based blind source separation in the analysis of long-duration field recordings. Our preliminary results suggest that NMF-based blind source separation can effectively recognize biological and non-biological sounds without any learning database. It can also accurately separate different vocalizing animals and improve acoustic-based biodiversity monitoring in a noisy environment.

ONLINE MODIFICATION OF THE METHOD OF X-MEDIUM ON THE BASIS OF ANSAMBLY OF SELORGANIZED MAP T. KOHONEN

The modified X-means method for clustering in the case when observations are sequentially fed to processing the proposed. This approach’s based on the ensemble of the clustering neural networks, proposed ensemble contains the T. Kohonen’s self-organizing maps. Each of the clustering neural networks consist of different number of neurons, where number of clusters is connected with the quality of there neurons. All ensemble members process information that siquentionally is fed to the system in the parallel mode. The effectiveness of clustering process is determined using Caliński-Harabasz index. The self-learning algorithm uses similarity measure of special type that. The feature of proposed method is absent of the competition step, i.e. neuron-winner is not determined. A number of experiments has been held in order to investigate the proposed system’s properties. Experimental results have proven the fact that the system under consideration could be used to solve a wide range of Data Mining tasks when data sets are processed in an online mode. The proposed ensemble system provides computational simplicity, and data sets are pro-cessed faster due to the possibility of parallel tuning.

A self-learning algorithm for coordinated control of rooftop units in small- and medium-sized commercial buildings

With the advent of the smart grid, demand response (DR) has been implemented in many electric utility control areas to reduce peak demand in buildings during grid stress conditions. However, small- and medium-sized commercial buildings typically do not deploy a building energy management (BEM) system due to high costs of commercially available solutions. Thus, their participation in DR events implies manual control and shutting down major building loads (e.g., air conditioning systems) without considering occupant comfort. With rapid development of Internet of Things (IoT) technologies, some cost-effective IoT-based BEM systems have become available. Based on such systems, this paper presents an algorithm to automatically coordinate the operation of rooftop units (RTUs) in small- and medium-sized commercial buildings, thereby meeting the specified power limit (kW) during a DR event while taking into account occupant comfort. The proposed algorithm has been designed to intelligently learn building thermal properties using coarse-grained indoor temperature data from thermostats, thus avoiding the deployment of sophisticated sensors network. A mixed-integer linear programming model has been utilized to determine an optimal RTU control strategy during a DR event. The peak load shedding performance of the proposed strategy has been tested in an office building in Blacksburg, VA, USA. The experimental result demonstrates that the building could achieve the required peak load reduction and the computation time required by the proposed algorithm is less than 5min. This implies that with the proposed algorithm a building is capable of responding to a DR signal with a short notice, providing valuable demand-side resources for electricity capacity and ancillary markets.

Audiovisual Narrative Creation and Creative Retrieval: How Searching for a Story Shapes the Story

Media professionals – such as news editors, image researchers, and documentary filmmakers - increasingly rely on online access to digital content within audiovisual archives to create narratives. Retrieving audiovisual sources therefore requires an in-depth knowledge of how to find sources digitally. These storytelling practices intertwine search technologies with the user’s ideas and production cultures. This paper presents qualitative research insights into how media professionals search in digital archives to create (trans)medial narratives, and uses the notion of creative retrieval to unravel the dynamics of audiovisual narrative production. Creative retrieval combines ideas about the effects of media convergence on media content, theories about serendipitous information retrieval, and studies of creativity to argue that retrieval practices of media professionals who create audiovisual narratives are governed by organizational, technological and content affordances and constraints. The paper furthermore exemplifies the first stage of an ongoing research project in which a user-centered design approach guides open source self-learning search algorithm development to support creative retrieval.

Living with an autonomous spatiotemporal home heating system: Exploration of the user experiences (UX) through a longitudinal technology intervention-based mixed-methods approach

Rising energy demands place pressure on domestic energy consumption, but savings can be delivered through home automation and engaging users with their heating and energy behaviours. The aim of this paper is to explore user experiences (UX) of living with an automated heating system regarding experiences of control, understanding of the system, emerging thermal behaviours, and interactions with the system as this area is not sufficiently researched in the existing homes setting through extended deployment. We present a longitudinal deployment of a quasi-autonomous spatiotemporal home heating system in three homes. Users were provided with a smartphone control application linked to a self-learning heating algorithm. Rich qualitative and quantitative data presented here enabled a holistic exploration of UX. The paper's contribution focuses on highlighting key aspects of UX living with an automated heating systems including (i) adoption of the control interface into the social context, (ii) how users' vigilance in maintaining preferred conditions prevailed as a better indicator of system over-ride than gross deviation from thermal comfort, (iii) limited but motivated proactivity in system-initiated communications as best strategy for soliciting user feedback when inference fails, and (iv) two main motivations for interacting with the interface – managing irregularities when absent from the house and maintaining immediate comfort, latter compromising of a checking behaviour that can transit to a system state alteration behaviour depending on mismatches. We conclude by highlighting the complex socio-technical context in which thermal decisions are made in a situated action manner, and by calling for a more holistic, UX-focused approach in the design of automated home systems involving user experiences.

Adaptive Dynamic Programming-Based Optimal Control Scheme for Energy Storage Systems With Solar Renewable Energy

In this paper, a novel optimal energy storage control scheme is investigated in smart grid environments with solar renewable energy. Based on the idea of adaptive dynamic programming (ADP), a self-learning algorithm is constructed to obtain the iterative control law sequence of the battery. Based on the data of the real-time electricity price (electricity rate in brief), the load demand (load in brief), and the solar renewable energy (solar energy in brief), the optimal performance index function, which minimizes the total electricity cost and simultaneously extends the battery's lifetime, is established. A new analysis method of the iterative ADP algorithm is developed to guarantee the convergence of the iterative value function to the optimum under iterative control law sequence for any time index in a period. Numerical results and comparisons are presented to illustrate the effectiveness of the developed algorithm.

A self-learning approach for optimal detailed scheduling of multi-product pipeline

Pipeline transportation is cost-optimal in refined product transportation. However, the optimization of multi-product pipeline scheduling is rather complicated due to multi-batch sequent transportation and multi-point delivery. Even though many scholars have conducted researches on the issue, there is hardly a model settling the discontinuous constraints in the model as a result of batch interface migration. Moreover, through investigation, there is no self-learning approach to pipeline scheduling optimization at present. This paper considers batch interface migration and divides the model into time nodes sequencing issue and a mixed-integer linear programming (MILP) model with the known time node sequence. And a self-learning approach is proposed through the combination of fuzzy clustering analysis and ant colony optimization (ACO). This algorithm is capable of self-learning, which greatly improves the calculation speed and efficiency. At last, a real pipeline case in China is presented as an example to illustrate the reliability and practicability of the proposed model. • A self-learning algorithm is proposed for detailed optimal scheduling for product oil pipeline. • Two novel discontinuous process constraints are taken into consideration. • This algorithm can improve calculation speed and efficiency by itself. • A real example with six cases is given to demonstrate the method’s practicality.

A self-learning algorithm for estimating solar heat gain and temperature changes in a single-Family residence

To develop effective control optimization strategies for managing residential electricity consumption in a smart grid environment, predictive algorithms are needed that are simple to implement, minimize custom configuration, and provide sufficient accuracy to enable meaningful control decisions. This paper presents a self‐learning algorithm for predicting indoor temperature in a single-family residence using a first-order lumped capacitance (1R1C) model that can be used to evaluate the consequences of energy saving or load shifting strategies on thermal comfort. The algorithm is formulated in such a way that key design details such as window size/configuration, thermal insulation, and airtightness (all major factors that affect heat loss and solar heat gain) are combined into effective parameters that can be learned from observation. This approach eliminates the need for a custom configuration for each residence.The model was validated using experimental data from the National Institute of Standards and Technology (NIST) Net-Zero Energy Residential Test Facility. It was found that with a simple temperature decay test to determine a thermal time constant and a seven-day sliding window of training data to account for seasonal variations in other parameters, the algorithm can reliably predict indoor temperatures for a 24h period using a solar irradiance forecast, an outdoor air temperature forecast, and heat pump output. The average root mean square temperature prediction error was found to be 2.2%.

A biological-inspired episodic cognitive map building framework for mobile robot navigation

This article proposes a self-learning method of robotic experience for building episodic cognitive map using biologically inspired episodic memory. The episodic cognitive map is used for robot navigation under uncertainty. Two main challenges which include high computational complexity and perceptual aliasing are addressed. The episodic memory-driving Markov decision process is proposed to simulate the organization of episodic memory by introducing neuron activation and stimulation mechanism. Episodic memory self-learning model and algorithm are presented for building the episodic cognitive map based on episodic memory-driving Markov decision process. Uncertain information is considered to improve mapping performance. The presented method can realize robotic memory real-time storage, incremental accumulation, integration and updating. Based on the episodic cognitive map, the predicted episodic trajectory can simply be computed by activation spreading of state neurons. The experimental results for a mobile...

Emergency materials transportation model in disasters based on dynamic programming and ant colony optimization

PurposeWhile scheduling and transporting emergency materials in disasters, the emergency materials and delivery vehicles are arriving at the distributing center constantly. Meanwhile, the information of the disaster reported to the government is updating continuously. Therefore, this paper aims to propose an approach to help the government make a transportation plan of vehicles in response to the disasters addressing the problem of material demand and vehicle amount continual alteration.Design/methodology/approachAfter elaborating the features and process of the emergency materials transportation, this paper proposes an emergency materials scheduling model in the case of material demand and vehicle amount continual alteration. To solve this model, the paper provides the vehicle transportation route allocation algorithm based on dynamic programming and the disaster area supply sequence self-learning algorithm based on ant colony optimization. Afterwards, the paper uses the model and the solution approach to computing the optimal transportation scheme of the food supply in Lushan earthquake in China.FindingsThe case study shows that the model and the solution approach proposed by this paper are valuable to make the emergency materials transportation scheme precise and efficient. The problem of material demand and vehicle amount changing continually during the process of the emergency materials transportation is solved promptly.Originality/valueThe model proposed by this paper improves the existing similar models in the following aspects: the model and the solution approach can not only solve the emergency materials transportation problem in the condition of varying demand and vehicle amount but also save much computing time; and the assumptions of this model are consistent with the actual situation of the emergency relief in disasters so that the model has a broad scope of application.

Economic load dispatch problem: quasi-oppositional self-learning TLBO algorithm

This paper proposes a meta-heuristic algorithm named as quasi-oppositional self-learning teacher-learner-based-optimization (QOSLTLBO) for solving non-convex economic load dispatch (ELD) problem. The ELD problem is an essential concern of power system and its main objective is to allocate optimal power generation to each generating unit so as to minimize the total cost of generation while satisfying all constraints available in the system. The problem considered in this paper is a non-convex quadratic generation cost of the units (with or without valve-point loading effects) with power balance and generation limits as the system constraints. This model of generation cost is a continuous model of the ELD problem. The proposed algorithm includes a quasi-oppositional approach for better initialization of population. A self-learning phase is added after teacher phase and learner phase of basic teacher-learner-based-optimization (TLBO) algorithm to improve the convergence rate. To prove the efficacy and robustness of proposed algorithm, it is applied to solve ELD problem on different standard IEEE generator systems and the results, thus obtained are compared with other state-of-art algorithms. The minimum total cost of generation in all the cases are obtained from the proposed algorithm which proves its effectiveness over others. The additional advantage of the proposed QOSLTLBO algorithm is that it is kept free from algorithm-specific parameters like basic TLBO.

Dynamic self-learning water-masking algorithm for AATSR, MERIS, and SPOT VEGETATION

Within the ESA CCI 'Fire Disturbance' project (Guenther et al., 2012), a dynamic self-learning water masking approach was developed for AATSR, MERIS-FR(S), MERIS-RR, and for SPOT VEGETATION (SPOT-VGT) data. The primary goal of the development was to find for all sensors a generic algorithm by combining static water masks on a global scale with a self-learning algorithm. Our approach results in the generation of a dynamic water mask which helps to distinguish burned areas from other dark areas as, e.g., cloud or topographic shadows or coniferous forests. The use of static water masks as training areas for the learning algorithm must take into account that small and shallow water bodies may change in time and that a precise geo-location of the static water mask and the scene under investigation is mandatory. The comparison of the water masks derived from all sensors for a region in Kazakhstan demonstrates the quality of the new dynamic water masks. In addition, the advantages to other water masking algorithms (MOD44W, Hansen_GFC or IDEPIX) are shown. Furthermore, the dynamic water masks of AATSR, MERIS and SPOT-VGT for the same region are presented and discussed together with the use of more detailed static water masks.

Using the driving behaviour to design a forward collision probability index

This paper uses the driving behaviour to design a Forward Collision Probability Index (FCPI) for alerting and to assist the driver to keep a safety driving distance for avoiding the forward collision accident in highway driving. We use the time-to-collision (TTC) as a main factor for calculating the FCPI. The index of FCPI is easy understanding for the driver even those who have no professional knowledge in vehicle technology. A self-learning algorithm for reducing the wrong warnings is presented for obtaining a suitable FCPI for the driver, which means that calculating the FCPI could meet each driver's behaviour. For the value of FCPI, value 0 is indicating the 0% probability of forward collision, and values 0.5 and 1 are indicating the 50% and 100% probabilities of forward collision, respectively. The experimental results guaranteed that the self-learning algorithm could figure out an optimal FCPI index for each driver to meet his/her driving behaviour and could reduce the wrong warnings.

Using the driving behaviour to design a forward collision probability index

This paper uses the driving behaviour to design a Forward Collision Probability Index (FCPI) for alerting and to assist the driver to keep a safety driving distance for avoiding the forward collision accident in highway driving. We use the time-to-collision (TTC) as a main factor for calculating the FCPI. The index of FCPI is easy understanding for the driver even those who have no professional knowledge in vehicle technology. A self-learning algorithm for reducing the wrong warnings is presented for obtaining a suitable FCPI for the driver, which means that calculating the FCPI could meet each driver's behaviour. For the value of FCPI, value 0 is indicating the 0% probability of forward collision, and values 0.5 and 1 are indicating the 50% and 100% probabilities of forward collision, respectively. The experimental results guaranteed that the self-learning algorithm could figure out an optimal FCPI index for each driver to meet his/her driving behaviour and could reduce the wrong warnings.