Daniela Pamplona


U2IS, ENSTA-Paristech

828 Boulevard des Maréchaux

91120 Palaiseau, FRANCE

I have graduated from Instituto Superior Tecnico, Portugal in Applied Mathematics and Computation both Bachelor and Master levels.

My master thesis was taken at the Vislab - Computer and Vision Lab, Institute for Systems and Robotics, with Alexandre Bernardino in space variant vision. We developed a new method to implement and manipulate foveated images, i.e. images with higher resolution in the center than in the periphery of the field of view (as in the primate eye, where retinal image has higher resolution in the fovea).

From 2009 to 2014, I was a PhD candidate at the Frankfurt Institute for Advanced Studies. I was working with Constantin Rothkopf  and Jochen Triesch on the modeling the variability of Retinal Ganglion Cells directly from the statistics of naturalistic images across the field of view.

From 2014 to 2016, I was a postdoc at the Biovision team, INRIA Sophia Antipolis, working with Pierre Kornprobst  and Bruno Cessac . We were working in methods for analysing (retinal) spiking data. At the single cell level, we have proposed a new stimulus for an efficient estimation of receptive fields, the shifted stimulus. At the population level, we worked in a method that assuming a leaky-integrate-and-fire model it learns the population connectivity from an empirical raster.

Since 2017, I am a postdoc at U2IS, ENSTA-ParisTech, working with Antoine Manzanera. We are working in life long learning of visual representations, particularly in the problems of incremental learning and curiosity.

We are looking for interns, master students and PhD students. If you would like to work with me and Antoine Manzanera, whether you have an idea or not, just send me an informal email with your research interests and your availability. A list of possible projects is available in the bottom of the website.

Research Interests

Unsupervised Learning, Incremental learning, probabilistic models, information theory, attention, front-vision, Retina, ecology, vision, sensory coding, embodyment, perception, intelligent systems, decision making.


Journal Publications

Pamplona, D.; Hilgen, G.; Hennig, M.; Cessac, B.; Sernagor, E.; Kornprobst P.; Large visual neuron assemblies receptive fields estimation using a super-resolution approach, under review

Pamplona, D.; Rothkopf, C. A.; All different all equal: ecology and the eye's optics explain P-ganglion cells' anisotropies, under preparation

Cessac, B; Kornprobst, P.; Kraria, S.; Nasser, H.; Pamplona, D.; Portelli, G.; Vieville T. PRANAS: A New Platform for Retinal Analysis and Simulation, Frontiers NeuroInformatics,2017

Hilgen, G.; Pirmoradian, S.; Pamplona, D.; Kornprobst, P.; Cessac, B.; Hennig, M.H.; Sernagor E.; Pan-retinal characterisation of Light Responses from Ganglion Cells in the Developing Mouse Retina,Scientific Reports, 2017

Pamplona, D.; Triesch, J.; Rothkopf,C. A. ; Power spectra of the natural input to the visual system, Vision Research, 2013 Download code


Conference Publications

Pamplona, D; Manzanera A.; Naturally Constrained Online Expectation Maximization,International Conference on Pattern Recognition (ICPR), 2020 Poster

Pamplona, D.; Bernardino, A.; Smooth Foveal Vision with Gaussian Receptive Fields, 9th IEEE - RAS International Conference on Humanoids Robots, 2009


Conference Abstracts

Cessac, B.; Kornprobst, P.; Kraria, S.; Nasser, H.; Pamplona, D.; Portelli, G.; Vieville T.; ENAS: A new software for spike train analysis and simulation, Bernstein Conference 2016

Hilgen, G.; Softley, S.; Pamplona, D.; Kornprobst, P.; Cessac, B.; Sernagor, E.; The effect of retinal GABA Depletion by Allylglycine on mouse retinal ganglion cell responses to light, European Retina Meeting, 2015

Pamplona, D.; Hilgen, G.; Cessac, B.; Sernagor, E.; Kornprobst, P.; A super-resolution approach for receptive fields estimation of neuronal ensembles, 24th Annual Computational Neuroscience Meeting (CNS), 2015

Pamplona, D.; Cessac, B.; Kornprobst, P.; Shifting stimulus for faster receptive fields estimation of ensembles of neurons, Computational and Systems Neuroscience (Cosyne), 2015

Pamplona, D.; Triesch, J.; Rothkopf,C.; Eye's imaging process explains ganglion cells anisotropies, Computational and Systems Neuroscience (Cosyne), 2013

Pamplona, D.; Triesch, J.; Rothkopf,C.; The statistics of looking: Deriving properties of retinal ganglion cells across the visual field, 12th Annual meeting of the Vision Sciences Society, 2012 (oral presentataion)

Pamplona, D.; Triesch, J.; Rothkopf,C.; Predicting Ganglion Cells Variability, Computational and Systems Neuroscience (Cosyne), 2011

Pamplona, D.; Triesch, J.; Rothkopf,C.; Edge and image statistics across the visual field, Bernstein Conference, 2011

Pamplona, D.; Weber, C.; Triesch J.;Foveation with optimized receptive fields, Bernstein Conference, 2009

Tushev G.; Liu, M.; Pamplona, D.; Bornschein, J.; Weber, C.; Triesch J.; Foveated Vision with FPGA Camera, Bernstein Conference, 2009 (demo)


Since 2017, I am teaching several classes, namely: Introduction to Matlab, Visual Perception and Learning, Probabilities and Statistics III, Neuro-computational Models of Vision.

Possible Projects

Title Topic Key words Students backgound Length
Incremental learning of images classes with Gaussian Mixture Models Computer Vision Incremental, unsupervised learning; image contexts Machine Learning; Computer Vision; Applied Math;Probabilistics and Statistics min: 10 weeks max 6months
Fisher info based gaze in natural images Eye movements; Computational Neuroscience Active vision; Fisher info Cognitive science, Vision,Machine Learning, min: 3 months, max: 1year
Truncated online Expectation Maximization Machine Learning Unsupervised online ML Applied Math, Machine Learning min: 10 weeks max 6months
Pan retinal analysis of the aging impact of mouse RFs Computational Neuroscience; Reverse correlation; data analysis; fitting; visualization Machine Learning; Biology; Neuroscience; min 10 weeks max 3 months
Design and set up of a robot with dual cameras Robotics, Middle ware
robotics, middle ware min:3month max :6months
On the consequences of Fisher Infomation Matrix diagonal approximation Machine Learning/ probabilistics FIM, computational time, approximation Math, ML, engeneering min:10 weeks max 6 months
Benchmarking state of the art methods for indoor classification Computer Vision image classification, indoor, house places Machine Learning; Computer Vision; Applied Mathematics; Probabilistics and Statistics min: 10 weeks max 6months