This page lists the talks of the joint CML journal club in Hamburg and Beijing.
Summer Semester 2017
Thursday, 27 July 2017, 17:00 (sharp), UKE, W34, Room 330
The focus paper for the meeting was proposed by Philipp Taesler and Samson Chien:
Lukasz Kaiser, Aidan N. Gomez, Noam Shazeer, Ashish Vaswani, Niki Parmar, Llion Jones, Jakob Uszkoreit (2017)
One Model To Learn Them All
- Thursday, 14 July 2017, 16:15 (s.t.), Informatikum, F-231, Vogt-Koelln-Str. 30
This week's paper about neural turing machines was proposed by Stefan Heinrich.
Alex Graves et al. (2016)
Hybrid computing using a neural network with dynamic external memory. Nature 538, Oct 2016.
- Thursday, 29 June 2017, 17:00 (s.t.), UKE, building W34, room 330
This week's paper was proposed by Christoph Korn:
David J. Heeger (2017)
Theory of cortical function, Proceedings of the National Academy of Sciences 114.8 (2017): 1773-1782 DOI: 10.1073/pnas.1619788114
- Thursday, 01 June 2017, 17:00 (s.t.), Informatikum Haus-F, Room F-132, Vogt-Koelln-Str. 30.
The paper up for discussion was proposed by Nicholas Katzakis:
Jacob, Robert JK, et al. (1994)
Integrality and separability of input devices.
ACM Transactions on Computer-Human Interaction (TOCHI) 1.1 (1994): 3-26. http://www.cs.tufts.edu/~jacob/papers/tochi.pdf
- Thursday, 04 May 2017, 17:00 (s.t.), Informatikum Haus-F, Room F-231, Vogt-Koelln-Str. 30.
This time we will have a "Special Issue" of the journal club, where we extend our very first topic of deep reinforcement learning to continuous deep reinforcement learning.
Lillicrap, T. P., Hunt, J. J., Pritzel, A., Heess, N., Erez, T., Tassa, Y., Wierstra, D. (2015).
Continuous control with deep reinforcement learning
arXiv preprint arXiv:1509.02971.
- Thursday, 06 April 2017, 17:00 (s.t.), Informatikum Haus-F, Room F-332, Vogt-Koelln-Str. 30.
This time we will look into the field of situated reference resolution. The paper was proposed by Özge Alacam:
Kennington, Casey, and David Schlangen (2017),
A simple generative model of incremental reference resolution for situated dialogue
Computer Speech & Language 41 (2017): 43-67. http://dx.doi.org/10.1016/j.csl.2016.04.002
Winter Semester 2016
- Thursday, 23 February 2017, 16:15, Informatikum Haus-F, Room F-231, Vogt-Koelln-Str. 30.
German Parisi proposed a continuation on the topic of the last meeting, a paper on neural modeling of the ventriloquism effect:
Elissa Magosso, Cristiano Cuppini, Mauro Ursino (2012),
A Neural Network Model of Ventriloquism Effect and Aftereffect
PLOS One, http://dx.doi.org/10.1371/journal.pone.0042503
- Thursday, 09 February 2017, 16:00, Biological Psychology, and Neuropsychology building, Room 200, Von-Melle-Park 11.
As a fresh start for this year, Jonathan Tong and German Parisi proposed to read and discuss a paper on audiovisual perception of space:
Brian Odegaard, David R. Wozny, Ladan Shams (2015)
Biases in Visual, Auditory, and Audiovisual Perception of Space
PLOS Computational Biology, http://dx.doi.org/10.1371/journal.pcbi.1004649
- Thursday, 01 December 2016, 16:15, Building W34, Room 307, UKE, Martinistr. 52
Sasa Redzepovic from the Institute for Systems Neuroscience UKE proposed the paper, so we will meet at UKE again:
Patrick McClure & Nikolaus Kriegeskorte,
Representational Distance Learning for Deep Neural Networks
Happy New Year and New Chinese Year!
- Thursday, 17 November 2016, 16:15, Building W34, Room 330, UKE, Martinistr. 52
The paper we are going to discuss was proposed by Christoph Korn:
Radoslaw Martin Cichy, Aditya Khosla, Dimitrios Pantazis, Antonio Torralba & Aude Oliva:
Comparison of deep neural networks to spatio-temporal cortical dynamics of human visual object recognition reveals hierarchical correspondence
Scientific Reports 6, Article number: 27755 (2016), doi:10.1038/srep27755
- Thursday, 03 November 2016, 16:15, Meeting Room B-331, Informatikum, Vogt-Kölln-Str. 30
The paper for this meeting was proposed by Sebastian Starke and deals with generative modelling of visual data using generative adversarial neural networks:
Jun-Yan Zhu, Philipp Krähenbühl, Eli Shechtman, and Alexei A. Efros: Generative Visual Manipulation on the Natural Image Manifold
European Conference on Computer Vision (ECCV). 2016. https://arxiv.org/pdf/1609.03552v2.pdf
- Thursday, 20 October 2016, 16:00, Meeting Room, Building N43, UKE
Adam H. Marblestone, Greg Wayne and Konrad P. Kording: Toward an Integration of Deep Learning and Neuroscience
Front. Comput. Neurosci., 14 September 2016. http://dx.doi.org/10.3389/fncom.2016.00094.
- Wednesday, 05 October 2016, 16:15, Meeting Room B-334, Informatikum, Vogt-Kölln-Str. 30
L. Pinto and A. Gupta:
Supersizing Self-supervision: Learning to Grasp from 50K Tries and 700 Robot Hours
IEEE ICRA (2016), https://ieeexplore.ieee.org/document/7487517
Summer Semester 2016
- Thursday, 04 August 2016, 17:00, Seminar Room, W34, 3. floor, UKE, Martinistr. 52
de-Wit, L., Alexander, D., Ekroll, V. & Wagemans, J.:
Is neuroimaging measuring information in the brain?
Psychonomic Bulletin & Review 1–14 (2016).
The article makes an interesting argument, how "information" can and should be defined regarding the analyses and results in neuroscience. What information is "decoded" from the brain and is it the same information the brain itself uses in it's downstream computations? Are observable patterns in brain activity the "representation" of information - and is information absent if matching patterns can't be detected with current methods? This might be an interesting starting point for a discussion about what to keep in mind when devising information processing models based on neuroscientific evidence.
- Wednesday, 29 June 2016, 16:00, Building W32, 3rd floor, UKE, Martinistr. 52
Jonas E, Kording KP (2016):
Could a neuroscientist understand a microprocessor?
bioRxiv XXI:1–5. http://dx.doi.org/10.1101/055624
Since we wanted to further examine overlaps between computer- and neuroscience, I'd like to suggest the article mentioned by Claus Hilgetag in the talks at the CML theory workshop. It might be a good basis for a discussion about better approaches to conduct "reverse engineering" on biological systems. Further, we could talk about ideas on how to use the insights from neuroscientific experiments (e.g. in computer science/robotics), although on their own these experiments often have rather limited explanatory power regarding the functional principle of the brain as a whole.
- Wednesday, 15 June 2016, 16:15, Informatikum, Room G-102
Mnih, Volodymyr, et al.:
Human-level control through deep reinforcement learning.
Nature 518.7540 (2015): 529-533. http://www.nature.com/nature/journal/v518/n7540/abs/nature14236.html
To fuel discussion, we'll announce a paper for each meeting that should be read beforehand and will be discussed there. Matthias Kerzel and I decided to start the club with a nature article that drew a lot of attention in the beginning of last year.