Machine Learning ...
... On any Project

Heart rate estimation via Photoplethysmography Fetal movement detection Deep Learning on Video Doublecompression Statistical modelling of high-frequency noise Imageforensics on Smart- phone pictures Automatic numberplate recogntion Document identification Prototyping image retrieval systems Academic Industrial Selection of machine learning projects

Projects

Academics

Expanding Machine Learning Knowledge ‌ ‌ ‌ ‌ ‌
Ph.D.
Digital cues in Multimedia Forensics
Deep-Learning [1,4]
Classification [1,2,3,4,5]
Discriminative Models [1,2,3,5]
Generative Models [4]
Temporal Data [1,4]
Categorical Data [2,3]
Outlier Detection [4]
Benchmarking [1,4,5]
Open-Set Problem [2]
Statistical Modeling [4]
Metric Learning and Manifolds [4]
Spliced image $$ \Downarrow \mathrm{d_{Mahalanobis}} (\vec{x};\vec{\mu},\Sigma) $$ Detected splice



Studies
Machine Learning on Biosignals
Classification [6,7]
Discriminative Models [7]
Regresion [6]
Temporal Data [6,7]
Digital Filters [6,7]
Wavelet Denoising [7]
Setup for study on fetal movement


Distinguished Paper Award -- Raid 2020

[1] Compression Forensics on High-Quality Video
[2] Towards Open-Set Forensic Source Grouping on JPEG Header Information
[3] Forensic Source Identification using JPEG Image Headers: The Case of Smartphones
[4] Residual-based forensic comparison of Video Sequences
[5] Camera Fingerprinting Authentication Revisited (Distinguished Paper Award)
[6] Detection of Fetal Kicks using Body-worn Accelerometers during Pregnancy
[7] Unobtrusive Heart Rate Estimation during Physical Excercise using Photoplethsmographic and Acceleration Data


Industries

Putting Knowledge into Productive Code ‌ ‌ ‌ ‌ ‌
Fraud-
detection
Supporting a renowned Insurance Company
Reseache & Development
Document Identification
Image Retrieval
Consistency Checks
Statistical Modeling
Examining Large Datasets
From Prototyping ...
... to Productive Code



Intern
Automatic Numberplate Recognition
Image Segmentation
Optical Character Recognition (OCR)
SVMs
Data Recording
Data Preparation
Prototyping

Presenting Knowledge

Talks

Public and Invited ones ‌ ‌ ‌ ‌ ‌
MPTR presenting
Title Venue
Digital Cues of Multimedia Forensics Doctoral Defense, FAU, 2020
Do modern smartphones automatically forge images? Lightning-Talks, FAU, 2020
Source-Identication of Images with help of JPEG Header files Cast-Workshop, Fraunhofer-Institut Darmstadt, 2019
The digital Fingerprint of a Camera Sensor Lange Nacht der Wissenschaft, FAU, 2019
How do machines actually learn? Philosophical-Seminar, FAU, 2017
Bildkompression -- Oder wie das Kamel durch ein Nadelöhr kommt E-Werk Erlangen, 2014, available online (in German)
Contact: patrick.mullan@fau.de