Bernhard Bermeitinger
Title
Dr.
Last Name
Bermeitinger
First name
Bernhard
Email
bernhard.bermeitinger@unisg.ch
ORCID
Phone
+41 71 224 79 17
6 results
Now showing 1 - 6 of 6
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, 425-P: Machine Learning Identifies Glypican 4 as Key Predictor of Five-Year Mortality in Heart Failure Patients with Prediabetes or Diabetes(2025-06-20) - Some of the metrics are blocked by yourconsent settings
Item type:Publication, 1386-P: Machine Learning Predicts T2DM Incidence Using Basic Clinical and Laboratory Parameters(2025-06-20) - Some of the metrics are blocked by yourconsent settings
Item type:Publication, 426-P: Predicting Coronary Stenoses Using Machine Learning to Reduce Unnecessary Angiographies(2025-06-20) - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Subset Pretraining for Enhancing Neural Network Training EfficiencyWe propose a novel alternative to traditional randomly sampled mini-batches for gradient computation: using a fixed subset for complete pretraining of a neural network model. This approach enables deterministic convergence instead of a merely probabilistic one, as proven by the stochastic approximation theory, whose prerequisites are frequently violated by popular optimization algorithms. The approach is justified by the hypothesis that the loss minimum of the training set can be expected to be well-approximated by the minima of its subsets. Such subset minima can be computed in a fraction of the time necessary for optimizing with the whole training set. They are also compatible with efficient second-order optimization methods, such as the conjugate gradient optimizer. These methods are particularly efficient in the convex environment of the loss minimum. The image classification datasets MNIST, CIFAR-10, and CIFAR-100, (optionally extended by augmentation of training data) test this hypothesis. The experiments confirm that the models achieve performance equivalent to that when trained with the conventional training scheme. In conclusion, if the overdetermination ratio for the given model and dataset sufficiently exceed unity, even small subsets are representative. This results in a possible reduction of the computing expense to a tenth or less. This paper is an extended version of Spörer et al. [13].Type:conference paperJournal:Communications in Computer and Information ScienceVolume:2703 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Training Neural Networks in Single vs Double PrecisionThe commitment to single-precision floating-point arithmetic is widespread in the deep learning community. To evaluate whether this commitment is justified, the influence of computing precision (single and double precision) on the optimization performance of the Conjugate Gradient (CG) method (a second-order optimization algorithm) and RMSprop (a first-order algorithm) has been investigated. Tests of neural networks with one to five fully connected hidden layers and moderate or strong nonlinearity with up to 4 million network parameters have been optimized for Mean Square Error (MSE). The training tasks have been set up so that their MSE minimum was known to be zero. Computing experiments have disclosed that single-precision can keep up (with superlinear convergence) with double-precision as long as line search finds an improvement. First-order methods such as RMSprop do not benefit from double precision. However, for moderately nonlinear tasks, CG is clearly superior. For strongly nonlinear tasks, both algorithm classes find only solutions fairly poor in terms of mean square error as related to the output variance. CG with double floating-point precision is superior whenever the solutions have the potential to be useful for the application goal.Type:conference paperJournal:14th International Conference on Knowledge Discovery and Information Retrieval - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Mathematical Foundations of Data Science(Springer Cham, 2023-03-13); ; ; This textbook aims to point out the most important principles of data analysis from the mathematical point of view. Specifically, it selected these questions for exploring: Which are the principles necessary to understand the implications of an application, and which are necessary to understand the conditions for the success of methods used? Theory is presented only to the degree necessary to apply it properly, striving for the balance between excessive complexity and oversimplification. Its primary focus is on principles crucial for application success. Topics and features: Focuses on approaches supported by mathematical arguments, rather than sole computing experiences Investigates conditions under which numerical algorithms used in data science operate, and what performance can be expected from them Considers key data science problems: problem formulation including optimality measure; learning and generalization in relationships to training set size and number of free parameters; and convergence of numerical algorithms Examines original mathematical disciplines (statistics, numerical mathematics, system theory) as they are specifically relevant to a given problem Addresses the trade-off between model size and volume of data available for its identification and its consequences for model parametrization Investigates the mathematical principles involves with natural language processing and computer vision Keeps subject coverage intentionally compact, focusing on key issues of each topic to encourage full comprehension of the entire book Although this core textbook aims directly at students of computer science and/or data science, it will be of real appeal, too, to researchers in the field who want to gain a proper understanding of the mathematical foundations “beyond” the sole computing experience.Type:bookVolume:1Scopus© Citations 8