Institute Homepage
Institute Homepage
DE
Sign In
Research
Research
Research
Research Overview
Publications
Departments
Empirical Inference
Haptic Intelligence
Perceiving Systems
Physical Intelligence
Robotic Materials
Social Foundations of Computation
Max Planck Research Groups
Intelligent Construction and Building Systems
Interactive Learning
Start-Up Teams
CELLnROLL
Research Groups
AI Safety and Alignment
Algorithms and Society
Biomimetic Materials and Machines
Computational Applied Mathematics & AI Lab
Cooperative Machine Intelligence for People-Aligned Safe Systems
Cyborg Robotics and Intelligent Sensing
Deep Models and Optimization
Embodied Social Interaction
Human Aspects of Machine Learning
Learning and Dynamical Systems
Mechanics for Intelligent Structures
Neuromechanics of Movement
Organizational Leadership and Diversity
Robotic Composites and Compositions
Robust Machine Learning
Safety- and Efficiency- aligned Learning
Science and Probabilistic Intelligence
Wild, Efficient, and Innovative AI
About Us
About us
People
Faculty
People Directory
Alumni Network
Contact
Management & Boards
Contact
Directions
Our Institute
About our institute
Campus Overview
Campus Facilities
Code of Conduct
Points of Contact for Employees
Our History
100/10 year anniversary
Career
Career
Career
Career
Open positions
Doctoral Programs
Doctoral Programs Overview
International Max Planck Research School for Intelligent Systems
Max Planck ETH Center for Learning Systems
ELLIS PhD Program
Max Planck Artificial Intelligence Network
Training
Internships
Planck Academy
Service
Service
Service Units
Service Overview
Administration
Scientific Coordination Office
IT Services
Building Services
Welcome Service
Central Scientific Facilities
Facilities Overview
Materials
Medical Systems
Optics and Sensing Laboratory
Robotics
Scientific Computing
Software Workshop
Workshops
Workshop Overview
Fine Mechanical Workshop
Glass Workshop
Central Mechanical Workshop
Mechatronics Workshop
Campus Services
Campus Facilities
Library
Max Planck House Tübingen
Impact
Impact
Impact
Impact Overview
Diversity, Equity, and Inclusion
Sustainability
Entrepreneurship & Innovation
Cooperation
Partnerships and Collaborations
Doctoral Programs
Partners and Initiatives
Cyber Valley
European Laboratory for Learning and Intelligent Systems
ELLIS Institute Tübingen
Tübingen AI Center
People
News
Events
Research
Research
Research
Research Overview
Publications
Departments
Empirical Inference
Haptic Intelligence
Perceiving Systems
Physical Intelligence
Robotic Materials
Social Foundations of Computation
Max Planck Research Groups
Intelligent Construction and Building Systems
Interactive Learning
Start-Up Teams
CELLnROLL
Research Groups
AI Safety and Alignment
Algorithms and Society
Biomimetic Materials and Machines
Computational Applied Mathematics & AI Lab
Cooperative Machine Intelligence for People-Aligned Safe Systems
Cyborg Robotics and Intelligent Sensing
Deep Models and Optimization
Embodied Social Interaction
Human Aspects of Machine Learning
Learning and Dynamical Systems
Mechanics for Intelligent Structures
Neuromechanics of Movement
Organizational Leadership and Diversity
Robotic Composites and Compositions
Robust Machine Learning
Safety- and Efficiency- aligned Learning
Science and Probabilistic Intelligence
Wild, Efficient, and Innovative AI
About Us
About us
People
Faculty
People Directory
Alumni Network
Contact
Management & Boards
Contact
Directions
Our Institute
About our institute
Campus Overview
Campus Facilities
Code of Conduct
Points of Contact for Employees
Our History
100/10 year anniversary
Career
Career
Career
Career
Open positions
Doctoral Programs
Doctoral Programs Overview
International Max Planck Research School for Intelligent Systems
Max Planck ETH Center for Learning Systems
ELLIS PhD Program
Max Planck Artificial Intelligence Network
Training
Internships
Planck Academy
Service
Service
Service Units
Service Overview
Administration
Scientific Coordination Office
IT Services
Building Services
Welcome Service
Central Scientific Facilities
Facilities Overview
Materials
Medical Systems
Optics and Sensing Laboratory
Robotics
Scientific Computing
Software Workshop
Workshops
Workshop Overview
Fine Mechanical Workshop
Glass Workshop
Central Mechanical Workshop
Mechatronics Workshop
Campus Services
Campus Facilities
Library
Max Planck House Tübingen
Impact
Impact
Impact
Impact Overview
Diversity, Equity, and Inclusion
Sustainability
Entrepreneurship & Innovation
Cooperation
Partnerships and Collaborations
Doctoral Programs
Partners and Initiatives
Cyber Valley
European Laboratory for Learning and Intelligent Systems
ELLIS Institute Tübingen
Tübingen AI Center
People
News
Events
Back
Sebastian Gerwinn
Note
: Sebastian Gerwinn has transitioned from the institute (Alumni).
Empirical Inference
Alumni
sebastian.gerwinn@tuebingen.mpg.de
Projects
Publications
Unüberwachtes Lernen Steuerbarer Filter
Obwohl sich die Pixel-Darstellung eines Bildes unter affinen Transformationen wie Translation, Rotation und Skalierung stark ändert, bleibt der Inhalt des Bildes weitgehend unverändert. Insbesondere, wenn sich die Änderungen eines D-dimensionalen Lichtintensitätsvektors durch eine einparametrige Lie-Gruppe beschreiben lassen, ist es möglich, eine verlustfreie Bildrepräsentation zu finden, bei der eine Komponente dem Transformationsparameter entspricht und die anderen (D-1) Komponenten invariant sind unter der Lie-Gruppentransformation. Um solche Bildrepräsentationen abzuleiten, konstruieren wir geeignete generative Modelle, mit denen Steuerbare Filter auf unüberwachte Weise gelernt werden können. Insbesondere haben wir zeigen können, dass es möglich ist mit einer anti-symmetrischen Variante der Kanonischen Korrelationsanalyse (CCA), eine vollständige Basis für 32x32 Bildausschnitte zu bestimmen, die sich aus rotationsinvarianten Steuerbaren Filtern zusammensetzt.
Bayesian Models for Multi-Electrode Neuronal Spike Recordings
We investigate Bayesian methods to predict responses from multiple retinal and LGN ganglion cells, conditioned on visual stimuli.
This website uses cookies to ensure you get the best experience.
Learn more
.
Accept