Thorsteinn Rögnvaldsson

Learn More
The Langevin updating rule, in which noise is added to the weights during learning , is presented and analyzed. It is well controlled and, being a natural extension to standard backpropagation learning, easily combined with other modiications of back-propagation. If the Hessian matrix is numerically ill-conditioned, Langevin updating converges faster than(More)
The discrimination powers of Multilayer perceptron (MLP) and Learning Vector Quantisation (LVQ) networks are compared for overlapping Gaussian distributions. It is shown, both analytically and with Monte Carlo studies, that the MLP network handles high dimensional problems in a more eecient way than LVQ. This is mainly due to the sigmoidal form of the MLP(More)
EKA Chemicals produces chemicals for pulp bleaching processes and paper making. The problem studied in the DALLAS project is the industrial production of hydrogen peroxide at EKA Chemicals' plant in Bohus, Sweden. Figure 4.1 shows a simple sketch of the hydrogen peroxide production process. A " working solution " , a mixture of a complex organic chemical(More)
Embedded network software has become increasingly interesting for both research and business as more and more networked embedded systems emerge. Well-known infrastructure protocol stacks are reimplemented on new emerging embedded hardware and software architectures. Also, newly designed or revised protocols are implemented in response to new application(More)
This paper presents an overview of an autonomous robotic material handling system. The goal of the system is to extend the functionalities of traditional AGVs to operate in highly dynamic environments. Traditionally, the reliable functioning of AGVs relies on the availability of adequate infrastructure to support navigation. In the target environments of(More)
  • 1