[AI Summary]: This talk presents research applying deep learning methods to automatically date Gottfried Wilhelm Leibniz’s handwritten manuscripts, addressing a major challenge in preparing his historical-critical edition. The project uses computer vision to detect fine-grained writing style evolution over Leibniz’s five-decade career, aiming for ±1 year accuracy without requiring transcription or OCR. The approach involves training models on ~600 pages of dated materials (1669-1716) using line-level segmentation and binarization to focus on handwriting features rather than material properties. This collaboration between BBAW’s Leibniz research group, TELOTA digital humanities team, and Hasso Plattner Institute demonstrates how machine learning can support traditional editorial scholarship.
The talk takes place on Monday, October 27, 2025 at 4:15 PM (16:00 c.t.) Berlin time via virtual conference room.
- Speakers: Tim Westphal (BBAW, TELOTA), Marco Santi (BBAW), Mario Tormo Romero (Hasso-Plattner-Institut, Potsdam)
- Format: Online (virtual)
- Language: German
- Registration: Not required
- Link: DH-Kolloquium BBAW
Overview
Part of the ongoing DH Colloquium series at the Berlin-Brandenburg Academy of Sciences (BBAW). The colloquium focuses on practical applications and critical reflection on digital humanities research.
- Series: Virtuelles DH-Kolloquium an der BBAW
- Host: BBAW Digital Humanities
- Website: DH-Kolloquium – Berlin-Brandenburgische Akademie der Wissenschaften