White Box vs. Black Box Modeling: On the Performance of Deep Learning, Random Forests, and Symbolic Regression in Solving Regression Problems

Affenzeller, Michael; Burlacu, Bogdan; Dorfer, Viktoria; Dorl, Sebastian; Halmerbauer, Gerhard; Königswieser, Tilman; Kommenda, Michael; Vetter, Julia; Winkler, Stephan

doi:10.1007/978-3-030-45093-9_35

Cited by 9 publications

(16 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A disadvantage of machine learning methods they often rely on specialized computer coding languages that are not always publicly available (Watson et al, 2019), although a number of open-source applications, including an R-package and a Python-based implementation, are openly available and easily accessible. In addition to the steep learning curve required to implement these methods, RF models are frequently referred to as "black box" methods, which implies that the internal algorithm decisions that produce the ultimate outcome are not always transparent, and it may be difficult to interpret the results (Affenzeller et al, 2020).…”

Section: Machine Learningmentioning

confidence: 99%

Surface monitoring of fire pollution

Bredder

2022

Preprint

View full text Add to dashboard Cite

This chapter discusses efforts to measure surface observations of air pollution at the country-scale. The countries with the most comprehensive regulatory systems to monitor air pollution are the older industrial nations such as countries in the United Kingdom and the United States. Recent proliferation of low-cost air quality monitors (LCAQM) are making near-real-time air pollution monitoring more prevalent across the globe. While unique challenges exist between regulatory and LCAQM data access and usability, there are common challenges in using these data for decision support and research applications. This chapter discusses common statistical methods for estimating air pollution including spatial interpolation methods, statistical regression methods, machine learning, and chemical transport modeling.

show abstract

Section: Machine Learningmentioning

confidence: 99%

Surface monitoring of fire pollution

Bredder

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…, 𝐸 𝑘 𝑛 𝑒 ), are constructed by maturation step 𝑘 from the elements of level 𝑘 − 1. However, due to boundary 1 ZGP is a proprietary algorithm with patent pending. An open source version is currently under development.…”

Section: E Zoetropic Representationmentioning

confidence: 99%

“…e performance of GPSR has been increased for instance by the combination of GP with more standard ML approaches [2]. Finally, novel benchmarks were established lately that also compare SR and classical ML algorithms on real datasets [26,38,1].…”

Section: Introductionmentioning

confidence: 99%

Zoetrope Genetic Programming for Regression

Boisbunon,

Fanara,

Grenet

et al. 2021

Preprint

View full text Add to dashboard Cite

e Zoetrope Genetic Programming (ZGP) algorithm is based on an original representation for mathematical expressions, targeting evolutionary symbolic regression.e zoetropic representation uses repeated fusion operations between partial expressions, starting from the terminal set. Repeated fusions within an individual gradually generate more complex expressions, ending up in what can be viewed as new features. ese features are then linearly combined to best t the training data. ZGP individuals then undergo speci c crossover and mutation operators, and selection takes place between parents and o spring. ZGP is validated using a large number of public domain regression datasets, and compared to other symbolic regression algorithms, as well as to traditional machine learning algorithms. ZGP reaches state-of-theart performance with respect to both types of algorithms, and demonstrates a low computational time compared to other symbolic regression approaches.

show abstract

“…In fact, to achieve good performance, RF usually combines at least 100 trees [26]. Therefore, RFs are commonly perceived as black box models [14,27,28], and various methods have been developed to interpret them [29,30]. However, these different approaches are not inherently incorporated into the RF's structure itself by design and need to be implemented as an additional layer applied after predictions.…”

Section: Introductionmentioning

confidence: 99%

Bayesian Networks in the Management of Hospital Admissions: A Comparison between Explainable AI and Black Box AI during the Pandemic

Nicora,

Catalano,

Bortolotto

et al. 2024

J. Imaging

View full text Add to dashboard Cite

Artificial Intelligence (AI) and Machine Learning (ML) approaches that could learn from large data sources have been identified as useful tools to support clinicians in their decisional process; AI and ML implementations have had a rapid acceleration during the recent COVID-19 pandemic. However, many ML classifiers are “black box” to the final user, since their underlying reasoning process is often obscure. Additionally, the performance of such models suffers from poor generalization ability in the presence of dataset shifts. Here, we present a comparison between an explainable-by-design (“white box”) model (Bayesian Network (BN)) versus a black box model (Random Forest), both studied with the aim of supporting clinicians of Policlinico San Matteo University Hospital in Pavia (Italy) during the triage of COVID-19 patients. Our aim is to evaluate whether the BN predictive performances are comparable with those of a widely used but less explainable ML model such as Random Forest and to test the generalization ability of the ML models across different waves of the pandemic.

show abstract

White Box vs. Black Box Modeling: On the Performance of Deep Learning, Random Forests, and Symbolic Regression in Solving Regression Problems

Cited by 9 publications

References 8 publications

Surface monitoring of fire pollution

Surface monitoring of fire pollution

Zoetrope Genetic Programming for Regression

Bayesian Networks in the Management of Hospital Admissions: A Comparison between Explainable AI and Black Box AI during the Pandemic

Contact Info

Product

Resources

About