R.S. Venema

Car license plate recognition with neural networks and fuzzy logic

A car license plate recognition system (CLPR-system) has been developed to identify vehicles by the contents of their license plate for speed-limit enforcement. This type of application puts high demands on the reliability of the CLPR-system. A combination of neural and fuzzy techniques is used to guarantee a very low error rate at an acceptable recognition rate. First experiments along highways in the Netherlands show that the system has an error rate, of 0.02% at a recognition rate of 98.51%. These results are also compared with other published CLPR-systems.

Learning feed-forward multi-nets

Multi-nets promise an improved performance over monolithic neural networks by virtue of their distributed implementation. This potential lacks popularity as, without precautions, the learning rate has to drop considerably to eliminate the occurrence of unlearning. This paper introduces extensions of the Error Back-Propagation algorithm to enable function preserving merging of neural modules at full learning rate.

Community-based observing networks and systems in the Arctic: Human perceptions of environmental change and instrument-derived data

Many papers have addressed the differing approaches to observation by scientists collecting instrumented data and by community or local knowledge-based observations. Integrating these ways of knowing is difficult because they operate at different scales and have different goals. It would benefit both scientists and communities to integrate community-based observations and instrumented data, despite obstacles, because it would expand scales of observation and because gauged data in the Arctic are sparse. This requires development of a protocol to integrate these knowledge systems to maximize reliability and validity. We used survey data from a community-based observing network in the Bering Sea and examined the correspondence of community-based observations with instrument-derived data for air temperature, sea ice break-up and freeze-up, and vegetation changes. Results highlight that there is a high correspondence between community-based observations for sea ice and vegetation change and instrumented data, but there is an inherent conflict in scales of observation for air temperature data. This helps to elucidate the benefits of community-based observing as a process for understanding and responding to change in the Arctic.

Growing filters for finite impulse response networks

Time-delay neural networks are well-suited for prediction purposes. A particular implementation is the finite impulse response (FIR) neural net. A major design problem exists in establishing the optimal order of such filters while minimizing the number of weights. Here, a constructive solution inspired by cascade learning is outlined and illustrated by some typical case-studies.

Invariant Subspaces and Regression

In linear regression a relationship between a response variable y and a set of predictor variables x 1,…, x p is modeled as y = β0 + β1 x 1 + … + β p x p + ε, where ε represents random error and the β i s are constant coefficients to be estimated from n observations of y and the x i . Given n observations of these p + 1 variables, the method of cyclic subspace regression (CSR) can be used to provide a finite number of estimators of β in the linear model y = X β + ε. Among the estimators produced by CSR are the estimators produced by the methods of least squares (LS), principal components regression (PCR), and partial least squares (PLS). In this article, after careful consideration of the invariant subspaces of X t X, a new method of regression is developed. This method, which the authors call invariant subspace regression (ISR), uses a selection of subspaces within the invariant subspaces of X t X, to create estimators of β. These subspaces are identified and created by the use of a finite list of user supplied non zero constants. Since there are no constraints on these constants other than they be non zero, ISR is capable of producing an infinite number of β estimators. It is shown that by an identified choice of constants all estimators produced by CSR can be generated by ISR, thereby showing that ISR is a generalization of CSR. Moreover, it is shown that all ISR estimators can be produced by CSR applied to an identified modification of the y data. Finally, examples are given showing that there exist ISR estimators that perform better in predictions than the best of all CSR estimators. It is also shown that there exist constants that create ISR estimators which perform equal to or worse than all CSR estimators in predictions. However, in this situation the best ISR estimator, in terms of prediction, is the best PCR estimator.

Flexible core generation for neural signal processing

The architecture and design of a CMOS core module for neural signal processing is presented. A logarithmic number representation is exploited to allow for multiplication with adaptive accuracy. This creates the facility to exchange accuracy for speed, both for the single instruction execution as during the parallelisation of a number of instructions. The core module is generated from a process independent silicon assembler. Details are given on a specific pipelined sign-magnitude implementation.

Designing ANN forecasting architectures from data conflict plots

One of the main issues in the analysis of a time series is its forecasting. Many questions arise in the design of a neural network that aims to capture the dynamics of a temporal sequence in order to predict it. In a reproducible way we want to find decision strategies for the preprocessing and the architecture of the network. In this paper we introduce a novel technique to extract important data features, called the data conflict plot. The conflict plot is used to design a modified architecture for the prediction of signals with distinct periodic components. Instead of a single delay line, this architecture is preceded by several incompletely connected delay lines.