Search

The availability of compact digital circuitry for the support of neural networks is a key requirement for resource-constrained embedded systems. This brief tackles the implementation of single hidden-layer feedforward neural networks, based on hard-limit activation functions, on reconfigurable devices. The resulting design strategy relies on a novel learning procedure that inherits the approach adopted in the Extreme Learning Machine paradigm. The eventual training process balances accuracy and network complexity effectively, thus supporting a digital architecture that prioritizes area utilization over computational performance. Experimental tests confirm that the design approach leads to efficient digital implementations of the predictor on low-performance devices. [ABSTRACT FROM AUTHOR]

The availability of compact fast circuitry for the support of artificial neural systems is a long-standing and critical requirement for many important applications. This brief addresses the implementation of the powerful extreme learning machine (ELM) model on reconfigurable digital hardware (HW). The design strategy first provides a training procedure for ELMs, which effectively trades off prediction accuracy and network complexity. This, in turn, facilitates the optimization of HW resources. Finally, this brief describes and analyzes two implementation approaches: one involving field-programmable gate array devices and one embedding low-cost low-performance devices such as complex programmable logic devices. Experimental results show that, in both cases, the design approach yields efficient digital architectures with satisfactory performances and limited costs. [ABSTRACT FROM AUTHOR]

The two major components of a robotic tactile-sensing system are the tactile-sensing hardware at the lower level and the computational/software tools at the higher level. Focusing on the latter, this research assesses the suitability of computational-intelligence (CI) tools for tactile-data processing. In this context, this paper addresses the classification of sensed object material from the recorded raw tactile data. For this purpose, three CI paradigms, namely, the support-vector machine (SVM), regularized least square (RLS), and regularized extreme learning machine (RELM), have been employed, and their performance is compared for the said task. The comparative analysis shows that SVM provides the best tradeoff between classification accuracy and computational complexity of the classification algorithm. Experimental results indicate that the CI tools are effective in dealing with the challenging problem of material classification. [ABSTRACT FROM AUTHOR]

A crucial issue in designing learning machines is to select the correct model parameters. When the number of available samples is small, theoretical sample-based generalization bounds can prove effective, provided that they are tight and track the validation error correctly. The maximal discrepancy (MD) approach is a very promising technique for model selection for support vector machines (SVM), and estimates a classifier's generalization performance by multiple training cycles on random labeled data. This paper presents a general method to compute the generalization bounds for SVMs, which is based on referring the SVM parameters to an unsupervised solution, and shows that such an approach yields tight bounds and attains effective model selection. When one estimates the generalization error, one uses an unsupervised reference to constrain the complexity of the learning machine, thereby possibly decreasing sharply the number of admissible hypothesis. Although the methodology has a general value, the method described in the paper adopts vector quantization (VQ) as a representation paradigm, and introduces a biased regularization approach in bound computation and learning. Experimental results validate the proposed method on complex real-world data sets. [ABSTRACT FROM AUTHOR]

A formal methodology drives the design and realization of a digital very large-scale Integration (VLSI) device supporting hierarchical vector equalization (HVQ) in computation-intensive co4ing applications. The hardware-oriented model-selection approach enhances the Minimum Description Length criterion with circuit-related aspects that allow consistent and efficient design. The resulting model parameters drive the subsequent realization in digital circuitry, which has first been implemented in field-programmable gate array (PA) technology to verify its correctness. The eventual VLSI realization results in an HVQ chip providing cost-effective, computationally chic lent real-time performances. Real-world applications support the consistency of the vector equalization approach and the effectiveness of the HVQ device. [ABSTRACT FROM AUTHOR]

Describes a methodology using circular backpropagation neural networks for the objective quality assessment of motion picture expert group video streams. Technical problem in digital television; Summary of the neural model adopted and the advantages in using the network for the specific multimedia application; Description of the neural-based system for video-quality evaluation.

Presents a study which examined the combination of the K-winner machine model with empirical measurements of a classifier's Vapnik-Chervonenkis-dim. Theoretical performance; Methods used; Findings and discussion.

Shows the design and prototypal hardware realization of a multiprocessor-oriented visual tracking system. Reliance of target identification on neural network's architecture; Industrial applications of the visual tracking of moving objects; Comparison with other approaches; Composition of the tracking mechanism; Techniques to approach tracking problems.

Presents information on a study which described an algorithm for training vector quantizers (VQ) with an improved version of the Neural Gas model and its implementation in analog circuitry. Theoretical aspects of VQ training; Circuit implementation of VQ training; Experimental results; Conclusions.