&lt;title&gt;Optimization and application of a RAM-based neural network for fast image processing tasks&lt;/title&gt;

Joergensen, Thomas Martini; Christensen, Steen Sloth; Andersen, Allan; Liisberg, C.

doi:10.1117/12.188904

Cited by 3 publications

(4 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As suggested in connection with equation (5) it can be advantageous to use individual values of the threshold level k forthe different classes. The challenge, however, is to fmd a design principle that is able to set the values properly.…”

Section: Selection Of the Threshold Level Kmentioning

confidence: 99%

See 1 more Smart Citation

<title>Cross-validation techniques for n-tuple-based neural networks</title>

Linneberg¹,

Jørgensen

1999

SPIE Proceedings

Self Cite

View full text Add to dashboard Cite

In spite of the simple classification concept, impressing performances have been reported using the n-tuple architecture in combination with a very simple training strategy. In general, however, the performance of the n-tuple classifier is highly dependent on the choice of input connections and on the encoding of the input data. Accordingly, the simple architecture needs to be accompanied with design tools for obtaining a suitable architecture. Due to the simplicity of the architecture, it is simple to perform leave-one-out cross-validation tests and extensions of the concept. Therefore, it is also possible to operate with design methods that make extensively use of such tests. This paper describes such design algorithms and especially introduces a simple design strategy that allows the n-tuple architecture to perform satisfactorily in cases with skewed class priors. It can also help to resolve conflicts in the training material. The described methods are evaluated on classification problems from the European StatLog project. It is hereby shown that the design tools extends the competitiveness of the n-tuple classification method.

show abstract

Section: Selection Of the Threshold Level Kmentioning

confidence: 99%

“…For all LUTs calculate their corresponding column entries, and set the output value of the target class to one in all the "active" columns. 5. for all examples in the training set.…”

Section: Tile Standard N-tuple Classifiermentioning

confidence: 99%

<title>Cross-validation techniques for n-tuple-based neural networks</title>

Linneberg¹,

Jørgensen

1999

SPIE Proceedings

Self Cite

View full text Add to dashboard Cite

show abstract

“…Estimating the generalization capabilities of a classifier can be done by performing a leave-one-out cross-validation where in turn a single example is left out of the training set; i.e., the classifier is trained on the reminder of the training set and tested on the one example excluded from the training set. As pointed out by Liisberg and described in [10] it is easy to incorporate a leave-one-out cross-validation test for the n-tuple classifier. The leave-one-out cross-validation classification of a training example x can be calculated as…”

Section: The N-tuple Classifiermentioning

confidence: 99%

“…A large average Hamming distance corresponds to a large variation or noise level (i.e., the probability, q c , of inverting the value of a bit) for the distribution in question. Expression (10) involves the calculation of overlap areas between different numbers of training examples and one test example. In order to average (10) over the training sets, the overlap areas must be replaced with the calculation of the average overlap areas between different numbers of training examples and a test example.…”

Section: Calculation Of Expected Output Scoresmentioning

confidence: 99%