Threshold Update Research Articles

An efficient online active learning algorithm for binary classification

Active learning is an important class of machine learning where labels are queried when necessary. Most active learning algorithms need to iteratively retrain the classifier when new labeled data are obtained. Such a batch learning process can incur a high overhead in both time and memory. In this paper, we propose a new online active learning algorithm for binary classification. Our algorithm uses the margin-based criterion, which compares the margin of instances with a threshold to decide whether it should be queried. Especially, we propose Iteratively Decreased Threshold (IDT), a new threshold update method for the margin-based criterion. By iteratively decreasing the threshold with IDT, our algorithm can effectively reduce the number of queried instances. In addition, as evaluating the margin-based criterion involves only simple inner productions, our algorithm is also very efficient to evaluate. We compare our algorithm with other state-of-the-art online active learning algorithms on six data sets, demonstrating that it requires less queries to achieve the same classification accuracy, and incurs a smaller computation overhead at the same time.

Fuzzy-neural self-adapting background modeling with automatic motion analysis for dynamic object detection

In this paper we propose a system that involves a Background Subtraction, BS, model implemented in a neural Self Organized Map with a Fuzzy Automatic Threshold Update that is robust to illumination changes and slight shadow problems. The system incorporates a scene analysis scheme to automatically update the Learning Rates values of the BS model considering three possible scene situations. In order to improve the identification of dynamic objects, an Optical Flow algorithm analyzes the dynamic regions detected by the BS model, whose identification was not complete because of camouflage issues, and it defines the complete object based on similar velocities and direction probabilities. These regions are then used as the input needed by a Matte algorithm that will improve the definition of the dynamic object by minimizing a cost function. Among the original contributions of this work are; an adapting fuzzy-neural segmentation model whose thresholds and learning rates are adapted automatically according to the changes in the video sequence and the automatic improvement on the segmentation results based on the Matte algorithm and Optical flow analysis. Findings demonstrate that the proposed system produces a competitive performance compared with state-of-the-art reported models by using BMC and Li databases.

An Efficient Trust Management Technique for Delay Tolerant Network

Delay Tolerant Networks (DTNs) have high end-to-end latency, which is often faces disconnection, and unreliable wireless connections. It does not mean a delay service instead DTNs provides a service where network imposes disruption or delay. It operates in challenged networks with extremely limited resources such as memory size, CPU processing power etc. This paper presents an efficient trust managing mechanism for providing secure environment. The proposed dynamic trust management protocol uses a dynamic threshold updating which overcomes the problems with time changing dynamic characteristics by dynamically updating the criteria in response to changing network conditions. This reduces overheads and increases the efficient use of routing network even in conditions change. Also the dynamic threshold update reduces the false detection probability of the malicious nodes. To show the effectiveness of the proposed system, a detailed simulation in the presence of selfish and malicious nodes is performed with ONE simulator. Finally a comparative analysis of our proposed routing with previous routing protocols is also performed. The results demonstrate that presented algorithm deals effectively with selfish behavior with providing significant gain on effective delivery ratio in trade off with message overhead and delay.

Cooperation and charity in spatial public goods game under different strategy update rules

Human cooperation can be influenced by other human behaviors and recent years have witnessed the flourishing of studying the coevolution of cooperation and punishment, yet the common behavior of charity is seldom considered in game-theoretical models. In this article, we investigate the coevolution of altruistic cooperation and egalitarian charity in spatial public goods game, by considering charity as the behavior of reducing inter-individual payoff differences. Our model is that, in each generation of the evolution, individuals play games first and accumulate payoff benefits, and then each egalitarian makes a charity donation by payoff transfer in its neighborhood. To study the individual-level evolutionary dynamics, we adopt different strategy update rules and investigate their effects on charity and cooperation. These rules can be classified into two global rules: random selection rule in which individuals randomly update strategies, and threshold selection rule where only those with payoffs below a threshold update strategies. Simulation results show that random selection enhances the cooperation level, while threshold selection lowers the threshold of the multiplication factor to maintain cooperation. When charity is considered, it is incapable in promoting cooperation under random selection, whereas it promotes cooperation under threshold selection. Interestingly, the evolution of charity strongly depends on the dispersion of payoff acquisitions of the population, which agrees with previous results. Our work may shed light on understanding human egalitarianism.

An improved pulse coupled neural network for image processing

To develop new image processing applications for pulse coupled neural network (PCNN), this paper proposes an improved PCNN model by redesigning the linking input, activity strength, linking weight, pulse threshold and pixel update rule. Two typical image processing examples based on such a model, namely fingerprint orientation field estimation and noise removal, are presented for explaining how to use the PCNN and determine parameters in image processing. Experiments show that the improved model is quite useful, and the PCNN-based approaches achieve better image processing results than the traditional ones.

Towards better making a decision in speaker verification

Speaker verification is a process that accepts or rejects the identity claim of a speaker. How to make a decision is a critical problem; a threshold for decision-making critically determines performance of a speaker verification system. Traditional threshold estimation methods take only information conveyed by training data into consideration and, to a great extent, do not relate it to production data. It turns out that a speaker verification system with such threshold estimation suffers from poor performance in reality due to mismatches. In this paper, we propose several methods towards better decision-making in a practical speaker verification system. Our methods include the use of additional reliable statistical information for threshold estimation, elimination of abnormal data for better estimation of underlying statistics, and on-line incremental threshold update. To evaluate the performance of our methods, we have done simulations based on a baseline system, Gaussian Mixture Model, in both text-dependent and text-independent modes. Comparative results show that in contrast to the recent threshold estimation methods our methods yield considerably better performance, especially on miscellaneous mismatch conditions, in terms of generalization. Thus our methods provide a promising way for real speaker verification applications.

Heuristic Methods for Dynamic Load Balancing in a Message-Passing Multicomputer

In this paper, a new adaptive scheme is presented for dynamic load balancing in a message-passing multicomputer. The scheme is based on using easy-to-implement heuristics and adaptive threshold in balancing the system load among dispersed nodes. It uses a distributed control over all computer nodes as coordinated by an information collector. Four heuristic methods are presented here, which are distinguished by the ranges for location and threshold update policies and by the disciplines used for determining the load transfer destination. A parallel simulator with distributed load balancers is developed on an iPSC/2 hypercube multi-computer. The load balancing scheme is evaluated on the basis of the effects of system utilization, load imbalance, communication and migration overhead, and multicomputer size. Relative merits of the four methods are revealed under various physical configurations of the multicomputer. The application potentials are discussed for parallel execution of AI-oriented programs and distributed semantic network data bases.