PubMed Health⌕ Search

Biomedical subjects

H Schwenk

Publications and source records attributed to H Schwenk.

8 recordsLinked to original sources
PubMed2000

Boosting neural networks.

Boosting is a general method for improving the performance of learning algorithms. A recently proposed boosting algorithm, AdaBoost, has been applied with great success to several benchmark machine learning problems using mainly decision trees as base classifiers. In this article we investigate whether AdaBoost also works as well with neural networks, and we discuss the advantages and drawbacks of different versions of the AdaBoost algorithm. In particular, we compare training methods based on sampling the training set and weighting the cost function. The results suggest that random resampling of the training data is not the main explanation of the success of the improvements brought by AdaBoost. This is in contrast to bagging, which directly aims at reducing variance and for which random resampling is essential to obtain the reduction in generalization error. Our system achieves about 1.4% error on a data set of on-line handwritten digits from more than 200 writers. A boosted multilayer network achieved 1.5% error on the UCI letters and 8.1% error on the UCI satellite data set, which is significantly better than boosted decision trees.

Algorithms↗
PubMed1998

The diabolo classifier

We present a new classification architecture based on autoassociative neural networks that are used to learn discriminant models of each class. The proposed architecture has several interesting properties with respect to other model-based classifiers like nearest-neighbors or radial basis functions: it has a low computational complexity and uses a compact distributed representation of the models. The classifier is also well suited for the incorporation of a priori knowledge by means of a problem-specific distance measure. In particular, we will show that tangent distance (Simard, Le Cun, & Denker, 1993) can be used to achieve transformation invariance during learning and recognition. We demonstrate the application of this classifier to optical character recognition, where it has achieved state-of-the-art results on several reference databases. Relations to other models, in particular those based on principal component analysis, are also discussed.

Journal Article↗
PubMed1994

[AnaTü-MidroTutor. An interactive tutoral program for a microscopy course].

The AnaTü-MikroTutor is an interactive tutorial program providing specific information for the microscopic anatomical course. It is offered as additional tool during practical microscopy and can be used for recapitulation of histology. The software is based on MS-DOS and can be implemented on IBM compatible computers. The main menu imitates a microscopic desk with a microscope, folders for microscopic slides and written additional information to each slide. The various functions are activated by mouse click over the respective icons. The program is offered to the students parallel to the microscopic course at 4 workstations and at a single terminal at the microscopic hall during practical microscopy. The main body of the program contains digital images of the microscopic slides of the course of microscopy in Tübingen. Each slide is represented by an overview and up to six magnifications. Legends are available as overlay. In addition, textual information is offered to each slide, which is intended to initiate further studies, explain specific termini or to indicate clinical relevance.

Anatomy↗
PubMed1987

[The menstrual cycle after hormonal growth inhibition in girls].

Up to 1979 growth was inhibited with hormones in 43 girls in puberty with constitutional gigantism. The average reduction in size as compared to the calculated final body height was 7.5 cm, treatment having been started at a skeletal age of 12.3 years and with a treatment duration of 1.4 years. In over 90% of the cases, the first spontaneous menstruation after termination of therapy occurred within three months. During the first two years, cycles were likewise regular in 90% of the cases, with a duration of 4 +/- 1 weeks. Ovarian function in adolescence was thus found not to be impaired, but rather to be more stable as compared to the total population. It may be concluded from this that the preconditions for future fertility can be regarded as favorable.

Adolescent↗