05 Fakultät Informatik, Elektrotechnik und Informationstechnik

Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6

Browse

Filters

2020 - 20233

Settings

Publication Type: search.filters.UBSTyp.ZeitschriftenartikelSubject: search.filters.subject.620Institute: search.filters.UBSInstitut.Institut für Parallele und Verteilte Systeme

Search Results

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Metrics and algorithms for locally fair and accurate classifications using ensembles
    (2022) Lässig, Nico; Oppold, Sarah; Herschel, Melanie
    To obtain accurate predictions of classifiers, model ensembles comprising multiple trained machine learning models are nowadays used. In particular, dynamic model ensembles pick the most accurate model for each query object, by applying the model that performed best on similar data. Dynamic model ensembles may however suffer, similarly to single machine learning models, from bias, which can eventually lead to unfair treatment of certain groups of a general population. To mitigate unfair classification, recent work has thus proposed fair model ensembles , that instead of focusing (solely) on accuracy also optimize global fairness . While such global fairness globally minimizes bias, imbalances may persist in different regions of the data, e.g., caused by some local bias maxima leading to local unfairness . Therefore, we extend our previous work by including a framework that bridges the gap between dynamic model ensembles and fair model ensembles. More precisely, we investigate the problem of devising locally fair and accurate dynamic model ensembles, which ultimately optimize for equal opportunity of similar subjects. We propose a general framework to perform this task and present several algorithms implementing the framework components. In this paper we also present a runtime-efficient framework adaptation that keeps the quality of the results on a similar level. Furthermore, new fairness metrics are presented as well as detailed informations about necessary data preparations. Our evaluation of the framework implementations and metrics shows that our approach outperforms the state-of-the art for different types and degrees of bias present in training data in terms of both local and global fairness, while reaching comparable accuracy.
  • Thumbnail Image
    ItemOpen Access
    Analyzing the influence of hyper-parameters and regularizers of topic modeling in terms of Renyi entropy
    (2020) Koltcov, Sergei; Ignatenko, Vera; Boukhers, Zeyd; Staab, Steffen
    Topic modeling is a popular technique for clustering large collections of text documents. A variety of different types of regularization is implemented in topic modeling. In this paper, we propose a novel approach for analyzing the influence of different regularization types on results of topic modeling. Based on Renyi entropy, this approach is inspired by the concepts from statistical physics, where an inferred topical structure of a collection can be considered an information statistical system residing in a non-equilibrium state. By testing our approach on four models-Probabilistic Latent Semantic Analysis (pLSA), Additive Regularization of Topic Models (BigARTM), Latent Dirichlet Allocation (LDA) with Gibbs sampling, LDA with variational inference (VLDA)-we, first of all, show that the minimum of Renyi entropy coincides with the “true” number of topics, as determined in two labelled collections. Simultaneously, we find that Hierarchical Dirichlet Process (HDP) model as a well-known approach for topic number optimization fails to detect such optimum. Next, we demonstrate that large values of the regularization coefficient in BigARTM significantly shift the minimum of entropy from the topic number optimum, which effect is not observed for hyper-parameters in LDA with Gibbs sampling. We conclude that regularization may introduce unpredictable distortions into topic models that need further research.
  • Thumbnail Image
    ItemOpen Access
    AssistML : an approach to manage, recommend and reuse ML solutions
    (2023) Villanueva Zacarias, Alejandro Gabriel; Reimann, Peter; Weber, Christian; Mitschang, Bernhard
    The adoption of machine learning (ML) in organizations is characterized by the use of multiple ML software components. When building ML systems out of these software components, citizen data scientists face practical requirements which go beyond the known challenges of ML, e. g.,  data engineering or parameter optimization. They are expected to quickly identify ML system options that strike a suitable trade-off across multiple performance criteria. These options also need to be understandable for non-technical users. Addressing these practical requirements represents a problem for citizen data scientists with limited ML experience. This calls for a concept to help them identify suitable ML software combinations. Related work, e. g.,  AutoML systems, are not responsive enough or cannot balance different performance criteria. This paper explains how AssistML, a novel concept to recommend ML solutions, i. e.,  software systems with ML models, can be used as an alternative for predictive use cases. Our concept collects and preprocesses metadata of existing ML solutions to quickly identify the ML solutions that can be reused in a new use case. We implement AssistML  and evaluate it with two exemplary use cases. Results show that AssistML can recommend ML solutions in line with users’ performance preferences in seconds. Compared to AutoML, AssistML offers citizen data scientists simpler, intuitively explained ML solutions in considerably less time. Moreover, these solutions perform similarly or even better than AutoML models.