Florian Krautli

Florian Krautli

Auto-generated profile, based on publicly available information

Florian Krautli

Max Planck Institute for the History of Science, Berlin, Germany

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 September 2020 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Verified
Journal Article
Open access
OEOliver Eberle
Oliver Eberle
JBJochen Buttner
Jochen Buttner
FKFlorian Krautli
Florian Krautli
KMKlaus-Robert Müller
Klaus-Robert Müller
MVMatteo Valleriani
Matteo Valleriani
GMGrégoire Montavon
Grégoire Montavon
Many learning algorithms such as kernel machines, nearest neighbors, clustering, or anomaly detection, are based on distances or similarities. Before similarities are used for training an actual machine learning model, we would like to verify that they are bound to meaningful patterns in the data. In this paper, we propose to make similarities interpretable by augmenting them with an explanation . We develop BiLRP, a scalable and theoretically founded method to systematically decompose the output of an already trained deep similarity model on pairs of input features. Our method can be expressed as a composition of LRP explanations, which were shown in previous works to scale to highly nonlinear models. Through an extensive set of experiments, we demonstrate that BiLRP robustly explains complex similarity models, e.g., built on VGG-16 deep neural network features. Additionally, we apply our method to an open problem in digital humanities: detailed assessment of similarity between historical documents, such as astronomical tables. Here again, BiLRP provides insight and brings verifiability into a highly engineered and problem-specific similarity model.

2 October 2021 in Interdisciplinary Science Reviews
Verified
Journal Article
Open access
SDStephen Boyd Davis
Stephen Boyd Davis
OVOlivia Vane
Olivia Vane
FKFlorian Kräutli
Florian Kräutli
Questions of trust are increasingly important in relation to data and its use. The authors focus on humanities data and its visualization, through analysis of their own recent projects with museums, archives and libraries internationally. Their account connects the specifics of hands-on digital humanities work to larger epistemological questions. They discuss the sources of potential mistrust, and examine how different expectations and assumptions emerge depending on the use and user of the data; they offer a simple schema through which the implications may be traced. It is argued that vital issues of trust can be engaged with through design, which, rather than being conceived as a cosmetic finish, is seen as contributing insights and questions that affect the whole process. The article concludes with recommendations intended to be useful in both theory and practice.