VS298: Unsolved Problems in Vision: Difference between revisions

From RedwoodCenter
Jump to navigationJump to search
No edit summary
No edit summary
 
(123 intermediate revisions by 2 users not shown)
Line 1: Line 1:
This seminar is about unsolved problems in vision.
One of the goals of vision science is to understand the nature of perception and its neural substrates. There are now many well established techniques and paradigms in both psychophysics and neuroscience to address problems in vision. However, knowing how to frame these questions for investigation is not necessarily obvious. Nervous systems present us with stunning complexity, and the purpose of perception itself is deeply mysterious. The goal of this seminar course is to step back and ask, what are the important problems that remain unsolved in vision research, and how should these be approached empirically? The course will consist of alternating weeks of discussion and guest lectures by vision scientists who will frame their views of the core unsolved problems. Interdisciplinary groups of students will devise a practical research plan to address an unsolved problem of their choice.


blah blah
'''Instructors''': [mailto:sklein@berkeley.edu Stan Klein], [mailto:feldman@icsi.berkeley.edu Jerry Feldman], [mailto:baolshausen@berkeley.edu Bruno Olshausen], and [mailto:karlzipser@berkeley.edu Karl Zipser]<br />
 
'''GSI''': [mailto:daniel.coates@berkeley.edu Dan Coates]
'''Instructor''': [mailto:baolshausen@berkeley.edu Bruno Olshausen]


'''Enrollment information''':
'''Enrollment information''':


VS 298 (section 4), 2 units<br />
VS 298 (section 2), 2 units<br />
CCN: 66489
CCN: 66478


'''Meeting time and place''':
'''Meeting time and place''':
 
Tuesday 6-8, 560 Evans (Redwood Center conference room)
Monday 6-8, Evans 560


'''Email list''':
'''Email list''':


nss2014@lists.berkeley.edu [https://calmail.berkeley.edu/manage/list/ subscribe]
*Seminar mailing list vs298_unsolved_problems_in_vision@lists.berkeley.edu [https://calmail.berkeley.edu/manage/list/listinfo_subscribe/vs298_unsolved_problems_in_vision@lists.berkeley.edu subscribe]
 
*Lecture series mailing list [https://calmail.berkeley.edu/manage/list/listinfo_subscribe/unsolved-problems@lists.berkeley.edu subscribe]
 
'''Readings''':
 
Books and review articles:
* [http://www.naturalimagestatistics.net Natural Image Statistics] by Hyvarinen, Hurri & Hoyer
* Olshausen BA & Lewicki MS (2013)  What natural scene statistics can tell us about cortical representation.  In:  The Cognitive Neurosciences V.  [https://www.dropbox.com/s/4fo1mkjb8u5gtcj/olshausen-lewicki-review.pdf paper]
* Geisler WS (2008) Visual perception and the statistical properties of natural scenes. Annual Review of Psychology [https://www.dropbox.com/s/sn4wigpi7to874u/geisler-review.pdf paper]


Weekly schedule:
'''Weekly schedule''':


{| class="wikitable", border="1"
{| class="wikitable", border="1"
Line 32: Line 23:
! Date
! Date
! Topic/Reading
! Topic/Reading
! Presenter
|-  
|- valign="top"
| Sept. 2
| Feb. 3
| Introduction
| '''Redundancy reduction, whitening, and power spectrum of natural images'''
* [http://www.trincoll.edu/depts/ecopsyc/perils/folder2/issues.html Gibson's unsolved problems]
* Barlow (1961):  Theory of redundancy reduction [https://www.dropbox.com/s/y7qgngopjrf5pvs/barlow61.pdf paper]
|-  
* Atick (1992):  Theory of whitening [https://www.dropbox.com/s/xr9k56bxwc4bj9d/atick-redlich92.pdf paper]  
| Sept. 9
* Field (1987):  1/f<sup>2</sup> power spectrum and sparse coding [https://www.dropbox.com/s/jenlrgxxvm4h539/field87.pdf paper]
| Methodology in vision science (Stan Klein)<br />
Additional reading:
*  Double-judgment psychophysics: problems and solutions [http://cornea.berkeley.edu/pubs/34.pdf pdf] Read pp 1560-1567  This will give a glimpse into some of the issues involved with the relationship between detecting and identifying an object. The second part of the paper is more complicated.
* Attneave (1954) - 'Some informational aspects of visual perception' [https://www.dropbox.com/s/z57af14zjomnuxe/attneave54.pdf paper]
* Measuring, estimating, and understanding the psychometric function: A commentary [http://cornea.berkeley.edu/pubs/01-11_PERCEPTION_AND_PSYCHOPHYSICS-Measuring_estimating_and_understanding_the_psychometric_function.pdf pdf] I (Stan Klein) was an editor of a special issue of "Perception & Psychophyics" and I wrote the summary article not only commenting on a number of the articles, but also trying to clarify some misunderstood aspects in the field.
* Laughlin (1981) - Histogram equalization of contrast response [https://www.dropbox.com/s/lhdd9dpz22nqa73/laughlin81.pdf paper]
* Psychophysics : A Practical Introduction [http://site.ebrary.com/lib/berkeley/docDetail.action?docID=10391609 site] This is the text by Kingdom and Prins that I've used when teaching psychophysics methods. I suggest reading Chapters 2 & 3. Some of the dichotomies in Chapter 2 are directly relevant to a number of unsolved problems in vision. Some might even be insoluble.
* Srinivasan (1982) - 'Predictive coding: a fresh view of inhibition in the retina' [https://www.dropbox.com/s/rkvl45yiapf4jfz/srinivasan-etal82.pdf paper]
 
* Switkes (1978) - Power spectrum of carpentered environments [https://www.dropbox.com/s/d14ktt9kgjimlx6/switkes-carpentered.pdf paper]
Marcus background discussion<br />
* Ruderman (1997) - Why are images 1/f<sup>2</sup>? [https://www.dropbox.com/s/clbwm63gkn036xq/ruderman-scaling.pdf paper]
* Beyond Hubel and Wiesel [http://vimeo.com/87683403 video]
* Torralba & Oliva (2003) - Power spectrum of natural image categories [https://www.dropbox.com/s/ybgzoc6egw99wee/torralba-oliva03.pdf paper]
* Selected chapters from the Algebraic Mind [http://www.psych.nyu.edu/gary/TAM/intro_section.html#introduction introduction] [http://www.psych.nyu.edu/gary/TAM/rules.html#rules_and_variables rules_and_variables]
|<br />
* New Yorker piece on deep nets [http://www.newyorker.com/news/news-desk/is-deep-learning-a-revolution-in-artificial-intelligence site]
Anthony DiFranco<br />
|-  
Dylan Paiton<br />
| Sept. 16
Michael Levy
| Evening seminar, focus on student projects: form groups and discuss proposal topics [https://www.dropbox.com/s/2onrsi6n6338bjb/4Dec2014%20Grant%20PDF.pdf?dl=0 grant_format]
|- valign="top"
| Feb. 10
|  '''Whitening in time and color; Robust coding''' <br />
* Dong & Atick (1995):  spatiotemporal power spectrum of natural movies [https://www.dropbox.com/s/npqsum1jx35so8p/dong-atick95.pdf paper]  
* Ruderman (1998):  statistics of cone responses [https://www.dropbox.com/s/qpo6v0viush5k0o/ruderman-color.pdf paper]  
* Karklin & Simoncelli (2012):  noisy population coding of natural images [https://www.dropbox.com/s/m4wc2c2v9p0vqyu/karklin-simoncelli12.pdf paper]
Additional reading:
* Dong & Atick (1995) - spatiotemporal decorrelation using lagged and non-lagged cells  [https://www.dropbox.com/s/lt13hc4mgcre5j3/dong-atick-decorrelation.pdf paper]
* Doi & Lewicki (2007) - A theory of retinal population coding [https://www.dropbox.com/s/kz0jfeiblcpha1e/doi-lewicki07.pdf paper]
| <br />
Chayut Thanapirom<br />
Michael Levy<br />
Yubei Chen
|-
|-
| Feb. 17
| Sept. 19 <br />
|  ** Holiday **
(Friday) 11:00 a.m., 5101 Tolman
|
| <br /> '''Gary Marcus lecture: Computational diversity and the mesoscale organization of the neocortex''' [https://archive.org/details/Redwood_Center_2014_09_19_Gary_Marcus video]
|- valign="top"
<br />
| Feb. 24
The human neocortex participates in a wide range of tasks, yet superficially appears to adhere to a relatively uniform six-layered architecture throughout its extent. For that reason, much research has been devoted to characterizing a single "canonical" cortical computation, repeated massively throughout the cortex, with differences between areas presumed to arise from their inputs and outputs rather than from “intrinsic” properties. There is as yet no consensus, however, about what such a canonical computation might be, little evidence that uniform systems can capture abstract and symbolic computation (e.g., language) and little contact between proposals for a single canonical circuit and complexities such as differential gene expression across cortex, or the diversity of neurons and synapse types. Here, we evaluate and synthesize diverse evidence for a different way of thinking about neocortical architecture, which we believe to be more compatible with evolutionary and developmental biology, as well as with the inherent diversity of cortical functions. In this conception, the cortex is composed of an array of reconfigurable computational blocks, each capable of performing a variety of distinct operations, and possibly evolved through duplication and divergence. The computation performed by each block depends on its internal configuration. Area-specific specialization arises as a function of differing configurations of the local logic blocks, area-specific long-range axonal projection patterns and area-specific properties of the input. This view provides a possible framework for integrating detailed knowledge of cortical microcircuitry with computational characterizations. With Adam Marblestone, MIT and Tom Dean, Google
|  '''Higher-order statistics and sensory coding''' <br />
* Barlow (1972): Sparse coding [https://www.dropbox.com/s/iqm62gnmje61y5c/barlow72.pdf paper]<br />
* Field (1994): What is the goal of sensory coding? [https://www.dropbox.com/s/ufvvznv0xx48ncf/field94.pdf paper]<br />
* Bell & Sejnowski (1995):  Independent component analysis. [https://www.dropbox.com/s/qre5l2y1i6p5rlp/bell-sejnowski95.pdf paper]<br />
Additional reading:
* Redlich (1993):  Redundancy Reduction as a Strategy for Unsupervised Learning.  [https://www.dropbox.com/s/f9kp7pwh9g4n0xx/redlich93.pdf paper]
* Baddeley (1996): Searching for filter with 'interesting' output distributions: An uninteresting direction to explore?  [https://www.dropbox.com/s/cioqfzgadb1fd5l/baddeley96.pdf paper]
* O'regan & Noe (2001): A sensorimotor account of vision and visual consciousness  [https://www.dropbox.com/s/9yg4je9wb1lohuq/oregan-noe01.pdf paper]
| <br />
Karl Zipser <br />
Michael Levy<br />
Mayur Mudigonda
|- valign="top"
| March 3
|  '''ICA and sparse coding of natural images''' <br />
* Bell & Sejnowski (1997): ICA of natural images [https://www.dropbox.com/s/n2y1fqf9zix5wfc/bell-sejnowski97.pdf paper]<br/>
* Olshausen & Field (1997):  Sparse coding of natural images [https://www.dropbox.com/s/np4kiyo2yfqtyuq/olshausen-field97.pdf paper]<br />
* van Hateren & Ruderman (1998), Olshausen (2003):  ICA/sparse coding of natural video [https://www.dropbox.com/s/jucelyqdkde23g9/olshausen-video03.pdf paper1], [https://www.dropbox.com/s/f3mxyw1sw0devb4/vanhateren-ruderman98.pdf paper2] <br />
Additional reading:
* Olshausen & Field (1996): simpler explanation of sparse coding [https://www.dropbox.com/s/wridvqn9fqalnn5/olshausen-field96.pdf paper]
| <br />
Mayur Mudigonda<br />
Zayd Enam<br />
Georgios Exarchakis
|-  
|-  
| March 11 **Tuesday**
| Sept. 23
'''Statistics of natural sound and auditory coding''' <br />
Marcus discussion (postponed) <br />
* Clark & Voss: '1/f noise and music' [https://www.dropbox.com/s/mfidr4wfsuppgp3/voss-clarke78.pdf paper]<br />
Feldman background discussion<br />
* Smith & Lewicki: sparse coding of natural sound [https://www.dropbox.com/s/o4so96di3fdkzu4/smith-lewick06.pdf paper]<br />
*NEW MATERIALs:
* Klein/Deweese:  ICA/sparse coding of spectrograms [https://www.dropbox.com/s/6txhh3y3xapvvci/klein-kording03.pdf paper1], [https://www.dropbox.com/s/damynt0ruugy1v3/carlson-deweese12.pdf paper2]
* Feldman, J. (2008). From molecule to metaphor: A neural theory of language. MIT press. (look at Chapters 1, 2, 9, and 26 before class) [http://site.ebrary.com/lib/berkeley/detail.action?docID=10173587 pdf]
| <br />
*OLD MATERIALS:
Tyler Lee<br />
*Feldman, J. (2013). The neural binding problem (s). Cognitive neurodynamics, 7(1), 1-11. [ftp://ftp.icsi.berkeley.edu/pub/feldman/binding.cody.pdf pdf]
Yubei Chen<br />
*Feldman, J. & Narayanan, S. (2014). Affordances, Actionability, and Simulation. Affordances Workshop, Robotics Science and Systems 2014, Berkeley, CA [ftp://ftp.icsi.berkeley.edu/pub/feldman/affordances.jf.pdf pdf]
TBD
* Feldman, J. (2010). Ecological expected utility and the mythical neural code. Cognitive neurodynamics, 4(1), 25-35. [ftp://ftp.icsi.berkeley.edu/pub/feldman/eeu.pdf pdf]
|- valign="top"
* F. T. Sommer: Neural oscillatons and synchrony as mechanisms for coding, communication and computation in the visual system. Chapter in: The New Visual Neurosciences, Eds.: Leo M. Chalupa and John S. Werner, MIT Press (2014) [http://mitpress.mit.edu/sites/default/files/titles/content/9780262019163_toc_0001.pdf pdf_contents]
| March 17
|-
| '''Higher-order group structure''' <br />
| Sept. 30
* Geisler: contour statistics [https://www.dropbox.com/s/2167nccf3pbhl48/geisler-etal01.pdf paper] <br />
|  Discuss student projects<br />
* Hyvarinen:  subspace ICA/topgraphic ICA [https://www.dropbox.com/s/322pakrrau9vl5g/hyvarinen2000.pdf paper1], [https://www.dropbox.com/s/zigl3gzursjmod1/hyvarinen-hoyer01.pdf paper2]<br />
|-
* Lyu & Simoncelli: radial Gaussianization [https://www.dropbox.com/s/j87vewntbg1rzdl/lyu-simoncelli09.pdf paper]<br />
| Wed. Oct. 1, 4:15 p.m., 489 Minor Hall
Additional reading:
| <br />'''Feldman lecture: The neural binding problem(s) and related mysteries''' [https://archive.org/details/uc_berkeley_vs298_2014_10_01_Jerry_Feldman video]
* Parent & Zucker (1989): Trace Inference, Curvature Consistency, and Curve Detection, [https://www.dropbox.com/s/9a56qnpgaq7h60n/parent-zucker89.pdf paper]<br />
As with many other “problems” in vision and cognitive science, “the binding problem” has been used to label a wide range of tasks of radically different behavioral and computational structure. These include a “hard” version that is currently intractable, a feature-binding variant that is productive routine science and a variable-binding case that is unsolved, but should be solvable. The talk will cover all these and some related problems that seem intractably hard as well as some that are unsolved, but are being approached with current and planned experiments.
* Field et al. (1993): Contour Integration by the Human Visual System: Evidence for a Local “Association Field” [https://www.dropbox.com/s/qgcsvzk2i2is5d0/field-etal93.pdf paper]<br />
<br />
* Zetzsche et al. (1999): The atoms of vision: Cartesian or polar? [https://www.dropbox.com/s/78th56ytm8rayjs/zetzsche-etal99.pdf paper]<br />
|-
* Garrigues & Olshausen (2010): Group Sparse Coding with a Laplacian Scale Mixture Prior, [https://www.dropbox.com/s/mgfg0rt7q9tokbz/garrigues-olshausen10.pdf paper]
| Oct. 7
| <br />
| Feldman discussion <br />
Chayut Thanapirom<br />
Malik background discussion<br />
Guy Isely<br />
* R. Girshick, J. Donahue, T. Darrell and J. Malik, "Rich feature hierarchies for accurate object detection and semantic segmentation". Proc. of CVPR 2014. [http://www.cs.berkeley.edu/~rbg/papers/r-cnn-cvpr.pdf pdf]
TBD
 
*P. Arbelaez, J. Pont-Tuset, J. Barron, F. Marques and J. Malik, "Multiscale Combinatorial Grouping", Proc. of CVPR 2014. [http://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/mcg/ pdf]
 
*B. Hariharan, P. Arbelaez, R. Girshick and J. Malik, Simultaneous Detection and Segmentation. ECCV (7) 2014: 297-312. [http://www.cs.berkeley.edu/~bharath2/pubs/pdfs/BharathECCV2014.pdf pdf]
 
*S. Gupta, R. Girshick, P. Arbelaez, J. Malik: Learning Rich Features from RGB-D Images for Object Detection and Segmentation. ECCV (7) 2014: 345-360. [http://www.cs.berkeley.edu/~sgupta/pdf/rcnn-depth.pdf pdf]
|-
|-
| March 24
|Tue. Oct. 14, 6 to 8 p.m. 560 Evans
|  ** Spring recess **
| <br />'''Jitendra Malik lecture: The Three R's of Computer Vision: Recognition, Reconstruction and Reorganization''' [https://archive.org/details/ucb_vs298_2014_10_14_Jitendra_Malik video]
|
<br />
|- valign="top"
| March 31
|  '''Energy-based models''' <br />
* Hinton:  Product of experts models, [https://www.dropbox.com/s/iow62b9nqbbxftw/hinton-poe-nc02.pdf paper] <br />
* Osindero & Hinton:  Product of Experts model of natural images, [https://www.dropbox.com/s/rkl97yw1ryvosya/osindero-welling-hinton06.pdf paper]<br />
* Roth & Black:  Fields of experts, [https://www.dropbox.com/s/w6lhf9li8tsexnm/roth-black05.pdf paper] <br />
Additional reading:
* Hinton:  Practical guide to training RBMs [https://www.dropbox.com/s/8kxbkrmay5h9abf/hinton-rbm-guideTR.pdf paper] <br />
* Teh et al:  Energy-based models for sparse overcomplete representation, [https://www.dropbox.com/s/ph17gczh2l1e9qa/teh-etal-jmlr03.pdf paper]<br />
* Zhu, Wu & Mumford: FRAME (Filters, random fields, and maximum entropy), [https://www.dropbox.com/s/fxcc1gx1vfz1mwo/zhu-wu-mumford-FRAME.pdf paper]
| <br />
Evan Shelhamer<br />
Brian Cheung<br />
Chris Warner
|-
|-
| April 7
| Oct. 20, 12-1:30 p.m., Minor 489
| '''Learning invariances through 'slow feature analysis'''' <br />
| <br />'''Harold Bedell lecture: Contour interaction: as far from the muddling crowd?''' [https://archive.org/details/ucb_vs298_2014_10_20_Harold_Bedell video]
* Foldiak/Wiskott: slow feature analysis, [https://www.dropbox.com/s/ce4ngfyop94whhj/foldiak91.pdf paper1], [https://www.dropbox.com/s/v3dgj50jp6sc26p/wiskott-sejnowski02.pdf paper2]<br />
Contour interaction describes the interference with target recognition that occurs in the presence of nearby flanking edges. As one of the pioneers in this research, Flom distinguished between contour interaction and crowding, in which contributions to spatial interference can derive also from additional factors, such as inaccurate eye movements and attentional processes. In the normal fovea, contour interaction and crowding have a similar magnitude and operate over a similar spatial extent. Both foveal contour interaction and crowding are reduced when the luminance of the stimulus is decreased. Unlike the fovea, the magnitude and extent of contour interaction in peripheral vision are considerably more limited than crowding. Further, peripheral contour interaction and crowding are not affected substantively by target luminance. Indeed, the magnitude and extent of peripheral contour interaction are similar for photopic and scotopic targets. These results suggest that the contributions of specific mechanisms may differ for foveal and peripheral contour interaction and crowding.
* Hyvarinen: 'Bubbles'  [https://www.dropbox.com/s/1u7cmflubvt0zfy/hyvarinen-bubbles03.pdf paper]<br />
<br />
* Berkes et al.: factorizing 'what' and 'where' from video, [https://www.dropbox.com/s/us4fmd6vphacc4x/berkes-etal09.pdf paper]
Readings:
| <br />
*Siderov, J., Waugh, S. J., & Bedell, H. E. (2013). Foveal contour interaction for low contrast acuity targets. Vision research, 77, 10-13. [https://www.dropbox.com/s/chxuluvmy1hrumh/Siderov_Bedell_Waugh2013.pdf?dl=0 pdf]
Guy Isely<br />
*Coates, D. R., & Levi, D. M. (2014). Contour interaction in foveal vision: A response to. Vision research, 96, 140-144. [https://www.dropbox.com/s/7tycw0m71ocw4cw/Coates%20Levi%20Ltr%20VR_14.pdf?dl=0 pdf]
Chayut Thanapirom<br />
*Siderov, J., Waugh, S. J., & Bedell, H. E. (2014). Foveal contour interaction on the edge: Response to ‘Letter-to-the-Editor’by Drs. Coates and Levi. Vision research, 96, 145-148. [https://www.dropbox.com/s/2so6z5yxrsgs5iy/Siderov%20etal%20CI%20on%20Edge%20Reply%20VR_14.pdf?dl=0 pdf]
Bharath Hariharan
 
|-
|-
| April 14
| Oct. 21
| '''Manifold and Lie group models''' <br />
| Malik discussion <br />
* Carlsson et al.:  Klein bottle model of natural images, [https://www.dropbox.com/s/egaeuqpr7spuaaq/carlsson-etal07.pdf paper]
Nakayama and Shimojo background discussion<br />
* Culpepper & Olshausen: Learning manifold transport operators, [https://www.dropbox.com/s/1yqnpg7mfodfd8y/culpepper-olshausen09.pdf paper]
* Nakayama, K. (1999). Mid-level vision. In R. A. Wilson & F. C. Keil (Eds.), The MIT encylopedia of the cognitive sciences Cambridge: MIT Press [http://visionlab.harvard.edu/members/ken/Papers/100mitencyclopedia99.pdf pdf]
* Roweis & Saul: Local Linear Embedding, [https://www.dropbox.com/s/xbslejj4jl7723q/roweis-saul00.pdf paper]
* Nakayama, K. (2010)  "Vision going social." The science of social vision. Adams, R.B. Jr., Ambady, N., Nakayama, K. & Shimojo, S. (Eds) Oxford University Press [http://visionlab.harvard.edu/members/ken/Papers/160NakayamaVisionGoingSocial.pdf pdf]
| <br />
* Nakayama, K. and Martini, P. (2011) Situating Visual Search. Vision Research, 51, 1526-1537. [http://visionlab.harvard.edu/members/ken/Papers/166NakayamaMartiniSVS2011.pdf pdf]
Yubei Chen<br />
(All Nakayama pubs available [http://visionlab.harvard.edu/members/ken/nakayamapub.htm here])
Bruno/Mayur<br />
*Shimojo, S. (2014). Postdiction: its implications on visual awareness, hindsight, and sense of agency. Frontiers in psychology, 5. [http://journal.frontiersin.org/Journal/10.3389/fpsyg.2014.00196/full pdf]
James Arnemann
|-
|-
| April 21
| Oct. 28, 6-8 p.m., 560 Evans
| '''Hierarchical models''' <br />
| <br />'''Ken Nakayama lecture: The scientist’s choice: solving, explaining, discovering . . . . '''
* Karklin & Lewicki (2003): density components, [https://www.dropbox.com/s/urjwi875vhtmyww/karklin-lewicki03.pdf paper] <br />
<br />
* Shan & Cottrell:  stacked ICA, [https://www.dropbox.com/s/vit1tyjsz75jalf/shan-cottrell07.pdf paper] <br />
* Cadieu & Olshausen (2012):  learning intermediate representations of form and motion, [https://www.dropbox.com/s/p0bsnjxq3v6rs0x/cadieu-olshausen12.pdf paper]
| <br />
Tyler Lee<br />
Brian Cheung<br />
Dylan Paiton
|-
|-
| April 28
| Nov. 3 (Monday)
| '''Deep network models''' <br />
12:00 p.m.
* Hinton & Salakhudinov (2006):  stacked RBMs, [https://www.dropbox.com/s/bjtfiiu44skwuzl/hinton-salakutdinov06.pdf paper] <br />
489 Minor Hall
* Le et al. (2011): Unsupervised learning (Google brain, 'cat' neurons), [https://www.dropbox.com/s/ydd91bhv0qj69rr/le-etal12.pdf paper] <br />
| <br />'''Shinsuke Shimojo lecture: Postdiction: its implications on visual awareness, hindsight, and sense of agency''' [https://archive.org/details/ucb_vs298_2014_11_03_Shinsuke_Shimojo video]
* Krishevsky et al. (2012): Supervised learning, ImageNet 1000 [http://books.nips.cc/papers/files/nips25/NIPS2012_0534.pdf paper]
<br />
| <br />
|- |-
TBD <br />
| Nov. 3 (Monday)
TBD <br />
3:30-4:30 p.m.
[mailto:abbasi@berkeley.edu Reza Abbasi-Asl]
560 Evans Hall
| <br />Discussion with Shinshuke Shimojo
<br />
|-  
| Nov. 4
| Nakayama and Shimojo discussion <br />
Wandell background discussion<br />
* To appear: Computational modeling of responses in human visual cortex. BA Wandell, J Winawer, KN Kay.
In Brain Mapping:  An Encyclopedic Reference (Edited by Thompson and Friston.) [http://white.stanford.edu/~brian/papers/mri/2014-WandellWinawerKay-CorticalModeling-Encyclopedia.pdf pdf]
|-
| (Friday) Nov. 14 11 a.m., 560 Evans Hall
| <br />'''Brian Wandell lecture'''
<br />
|-
| Nov. 18
| Consciousness discussion <br />
 
|-
|-
| May 6 <br />
| Nov. 25
Note: Tuesday
|| <br />
|  '''Special topics''' <br />
Koch background discussion<br />
* Fergus (2013):  visualizing what deep nets learn [http://arxiv.org/abs/1311.2901 paper]<br />
*Koch, C. Project MindScope [https://www.dropbox.com/s/zslgzwyu9pey1mz/Future%20of%20the%20Brain%20-%20MindScope%20%2715.pdf?dl=0 pdf]
* Schmidhuber: deep nets ([http://arxiv.org/pdf/1312.5548.pdf paper]), focusing on LOCOCODE ([http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.51.1412 paper])
*Tsuchiya, N., & Koch, C. (2008). The relationship between consciousness and attention. The neurology of consciousness: Cognitive neuroscience and neuropathology, 63-78. [https://www.dropbox.com/s/huehdgxd77jzjha/Tsuchiya%20%26%20Koch%20%2714.pdf?dl=0 pdf]
* Image compression with Hopfield networks <br />
*Klein, S. A. (1993). Will robots see? Chapter in Spatial Vision in Humans and Robots, Cambridge University Press, 184-199. [https://www.dropbox.com/s/0u15k6yf9o8zrvo/Klein%20Will%20Robots%20See.pdf?dl=0 pdf]
| <br />
*Tononi, G., & Koch, C. (2014). Consciousness: Here, There but Not Everywhere. arXiv preprint arXiv:1405.7089. [http://arxiv.org/pdf/1405.7089.pdf pdf]
[mailto:shiry@berkeley.edu Shiry Ginosar]<br />
*Scientific American [http://blogs.scientificamerican.com/cross-check/2012/04/02/christof-koch-on-free-will-the-singularity-and-the-quest-to-crack-consciousness/ article]
Anthony DiFranco<br />
*Scientific American [http://www.scientificamerican.com/article/consciousness-does-not-reside-here/ article]
Chris Hillar
|-
|-
| May 12
| Dec. 2, 4-6 p.m., 125 Li Ka Shing
| '''Special topics''' <br />
| <br />'''Christof Koch lecture: Unsolved Problems in Vision: Consciousness.'''
 
Evening seminar: Koch Discussion<br /><br />
|
|}
|}
<br />
'''Additional Materials'''
*recent special issue of CurrOpinNeuro [http://www.sciencedirect.com/science/journal/09594388/25/supp/C journal]
*Olshausen BA Olshausen (2013) Perception as an Inference Problem. [http://redwood.berkeley.edu/bruno/papers/perception-as-inference.pdf pdf]
*Olshausen BA (2012)  20 years of learning about vision: Questions answered, questions unanswered, and questions not yet asked. In: 20 Years of Computational Neuroscience (Symposium of the CNS 2010 annual meeting) [http://redwood.berkeley.edu/bruno/papers/CNS2010-chapter.pdf pdf]
*Kitaoka, A (2014) Color-dependent motion illusions in stationary images and their phenomenal dimorphism. Perception advance online publication [http://www.perceptionweb.com/openaccess/p7706.pdf pdf]
*O'Regan, J. K., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and brain sciences, 24(05), 939-973.[https://www.dropbox.com/s/tbih5x5vo1o5en7/2001%20A%20sensorimotor%20account%20of%20vision%20and%20visual%20consciousness.pdf?dl=0 pdf]
*Bruno Olshausen lecture (1 July 2014) 20 Years of Learning About Vision:  Questions Answered, Questions Unanswered, and Questions Not Yet Asked  [https://archive.org/details/ucb_vision_science_2014_07_01_Bruno_Olshausen video]
*Solari, S. V. H., & Stoner, R. (2011). Cognitive consilience: primate non-primary neuroanatomical circuits underlying cognition. Frontiers in neuroanatomy, 5. [http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3243081/pdf/fnana-05-00065.pdf pdf]
*Dyson, Freeman. The Case for Blunders. The New York Review of Books, 6 March 2014. [http://www.nybooks.com/articles/archives/2014/mar/06/darwin-einstein-case-for-blunders/ pdf]
*Machine-Learning Maestro Michael Jordan on the Delusions of Big Data and Other Huge Engineering Efforts,  20 Oct 2014, Lee Gomes, IEEE Specturm [http://spectrum.ieee.org/robotics/artificial-intelligence/machinelearning-maestro-michael-jordan-on-the-delusions-of-big-data-and-other-huge-engineering-efforts link]
*Yann LeCunn responds to Mike Jordan's Spectrum interview [https://www.facebook.com/yann.lecun/posts/10152348155137143 link]
*Kravitz, D. J., Saleem, K. S., Baker, C. I., Ungerleider, L. G., & Mishkin, M. (2013). The ventral visual pathway: an expanded neural framework for the processing of object quality. Trends in cognitive sciences, 17(1), 26-49. [http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3532569/pdf/nihms419686.pdf pdf]
* Vinyals, O. et al. Show and Tell: A Neural Image Caption Generator. 2014 arXiv.1411.4555v1 [http://arxiv.org/pdf/1411.4555v1.pdf pdf]
*Koch, C., & Tononi, G. (2011). A test for consciousness. Scientific American, 304(6), 44-47. [pdf https://www.dropbox.com/s/h2bo3swrjr1g1l1/A_Test_for_Consciousness.pdf?dl=0]

Latest revision as of 20:35, 4 December 2014

One of the goals of vision science is to understand the nature of perception and its neural substrates. There are now many well established techniques and paradigms in both psychophysics and neuroscience to address problems in vision. However, knowing how to frame these questions for investigation is not necessarily obvious. Nervous systems present us with stunning complexity, and the purpose of perception itself is deeply mysterious. The goal of this seminar course is to step back and ask, what are the important problems that remain unsolved in vision research, and how should these be approached empirically? The course will consist of alternating weeks of discussion and guest lectures by vision scientists who will frame their views of the core unsolved problems. Interdisciplinary groups of students will devise a practical research plan to address an unsolved problem of their choice.

Instructors: Stan Klein, Jerry Feldman, Bruno Olshausen, and Karl Zipser
GSI: Dan Coates

Enrollment information:

VS 298 (section 2), 2 units
CCN: 66478

Meeting time and place: Tuesday 6-8, 560 Evans (Redwood Center conference room)

Email list:

  • Seminar mailing list vs298_unsolved_problems_in_vision@lists.berkeley.edu subscribe
  • Lecture series mailing list subscribe

Weekly schedule:

Date Topic/Reading
Sept. 2 Introduction
Sept. 9 Methodology in vision science (Stan Klein)
  • Double-judgment psychophysics: problems and solutions pdf Read pp 1560-1567 This will give a glimpse into some of the issues involved with the relationship between detecting and identifying an object. The second part of the paper is more complicated.
  • Measuring, estimating, and understanding the psychometric function: A commentary pdf I (Stan Klein) was an editor of a special issue of "Perception & Psychophyics" and I wrote the summary article not only commenting on a number of the articles, but also trying to clarify some misunderstood aspects in the field.
  • Psychophysics : A Practical Introduction site This is the text by Kingdom and Prins that I've used when teaching psychophysics methods. I suggest reading Chapters 2 & 3. Some of the dichotomies in Chapter 2 are directly relevant to a number of unsolved problems in vision. Some might even be insoluble.

Marcus background discussion

Sept. 16 Evening seminar, focus on student projects: form groups and discuss proposal topics grant_format
Sept. 19

(Friday) 11:00 a.m., 5101 Tolman


Gary Marcus lecture: Computational diversity and the mesoscale organization of the neocortex video


The human neocortex participates in a wide range of tasks, yet superficially appears to adhere to a relatively uniform six-layered architecture throughout its extent. For that reason, much research has been devoted to characterizing a single "canonical" cortical computation, repeated massively throughout the cortex, with differences between areas presumed to arise from their inputs and outputs rather than from “intrinsic” properties. There is as yet no consensus, however, about what such a canonical computation might be, little evidence that uniform systems can capture abstract and symbolic computation (e.g., language) and little contact between proposals for a single canonical circuit and complexities such as differential gene expression across cortex, or the diversity of neurons and synapse types. Here, we evaluate and synthesize diverse evidence for a different way of thinking about neocortical architecture, which we believe to be more compatible with evolutionary and developmental biology, as well as with the inherent diversity of cortical functions. In this conception, the cortex is composed of an array of reconfigurable computational blocks, each capable of performing a variety of distinct operations, and possibly evolved through duplication and divergence. The computation performed by each block depends on its internal configuration. Area-specific specialization arises as a function of differing configurations of the local logic blocks, area-specific long-range axonal projection patterns and area-specific properties of the input. This view provides a possible framework for integrating detailed knowledge of cortical microcircuitry with computational characterizations. With Adam Marblestone, MIT and Tom Dean, Google

Sept. 23 Marcus discussion (postponed)

Feldman background discussion

  • NEW MATERIALs:
  • Feldman, J. (2008). From molecule to metaphor: A neural theory of language. MIT press. (look at Chapters 1, 2, 9, and 26 before class) pdf
  • OLD MATERIALS:
  • Feldman, J. (2013). The neural binding problem (s). Cognitive neurodynamics, 7(1), 1-11. pdf
  • Feldman, J. & Narayanan, S. (2014). Affordances, Actionability, and Simulation. Affordances Workshop, Robotics Science and Systems 2014, Berkeley, CA pdf
  • Feldman, J. (2010). Ecological expected utility and the mythical neural code. Cognitive neurodynamics, 4(1), 25-35. pdf
  • F. T. Sommer: Neural oscillatons and synchrony as mechanisms for coding, communication and computation in the visual system. Chapter in: The New Visual Neurosciences, Eds.: Leo M. Chalupa and John S. Werner, MIT Press (2014) pdf_contents
Sept. 30 Discuss student projects
Wed. Oct. 1, 4:15 p.m., 489 Minor Hall
Feldman lecture: The neural binding problem(s) and related mysteries video

As with many other “problems” in vision and cognitive science, “the binding problem” has been used to label a wide range of tasks of radically different behavioral and computational structure. These include a “hard” version that is currently intractable, a feature-binding variant that is productive routine science and a variable-binding case that is unsolved, but should be solvable. The talk will cover all these and some related problems that seem intractably hard as well as some that are unsolved, but are being approached with current and planned experiments.

Oct. 7 Feldman discussion

Malik background discussion

  • R. Girshick, J. Donahue, T. Darrell and J. Malik, "Rich feature hierarchies for accurate object detection and semantic segmentation". Proc. of CVPR 2014. pdf
  • P. Arbelaez, J. Pont-Tuset, J. Barron, F. Marques and J. Malik, "Multiscale Combinatorial Grouping", Proc. of CVPR 2014. pdf
  • B. Hariharan, P. Arbelaez, R. Girshick and J. Malik, Simultaneous Detection and Segmentation. ECCV (7) 2014: 297-312. pdf
  • S. Gupta, R. Girshick, P. Arbelaez, J. Malik: Learning Rich Features from RGB-D Images for Object Detection and Segmentation. ECCV (7) 2014: 345-360. pdf
Tue. Oct. 14, 6 to 8 p.m. 560 Evans
Jitendra Malik lecture: The Three R's of Computer Vision: Recognition, Reconstruction and Reorganization video


Oct. 20, 12-1:30 p.m., Minor 489
Harold Bedell lecture: Contour interaction: as far from the muddling crowd? video

Contour interaction describes the interference with target recognition that occurs in the presence of nearby flanking edges. As one of the pioneers in this research, Flom distinguished between contour interaction and crowding, in which contributions to spatial interference can derive also from additional factors, such as inaccurate eye movements and attentional processes. In the normal fovea, contour interaction and crowding have a similar magnitude and operate over a similar spatial extent. Both foveal contour interaction and crowding are reduced when the luminance of the stimulus is decreased. Unlike the fovea, the magnitude and extent of contour interaction in peripheral vision are considerably more limited than crowding. Further, peripheral contour interaction and crowding are not affected substantively by target luminance. Indeed, the magnitude and extent of peripheral contour interaction are similar for photopic and scotopic targets. These results suggest that the contributions of specific mechanisms may differ for foveal and peripheral contour interaction and crowding.
Readings:

  • Siderov, J., Waugh, S. J., & Bedell, H. E. (2013). Foveal contour interaction for low contrast acuity targets. Vision research, 77, 10-13. pdf
  • Coates, D. R., & Levi, D. M. (2014). Contour interaction in foveal vision: A response to. Vision research, 96, 140-144. pdf
  • Siderov, J., Waugh, S. J., & Bedell, H. E. (2014). Foveal contour interaction on the edge: Response to ‘Letter-to-the-Editor’by Drs. Coates and Levi. Vision research, 96, 145-148. pdf
Oct. 21 Malik discussion

Nakayama and Shimojo background discussion

  • Nakayama, K. (1999). Mid-level vision. In R. A. Wilson & F. C. Keil (Eds.), The MIT encylopedia of the cognitive sciences Cambridge: MIT Press pdf
  • Nakayama, K. (2010) "Vision going social." The science of social vision. Adams, R.B. Jr., Ambady, N., Nakayama, K. & Shimojo, S. (Eds) Oxford University Press pdf
  • Nakayama, K. and Martini, P. (2011) Situating Visual Search. Vision Research, 51, 1526-1537. pdf

(All Nakayama pubs available here)

  • Shimojo, S. (2014). Postdiction: its implications on visual awareness, hindsight, and sense of agency. Frontiers in psychology, 5. pdf
Oct. 28, 6-8 p.m., 560 Evans
Ken Nakayama lecture: The scientist’s choice: solving, explaining, discovering . . . .


Nov. 3 (Monday)

12:00 p.m. 489 Minor Hall


Shinsuke Shimojo lecture: Postdiction: its implications on visual awareness, hindsight, and sense of agency video


Nov. 3 (Monday)

3:30-4:30 p.m. 560 Evans Hall


Discussion with Shinshuke Shimojo


Nov. 4 Nakayama and Shimojo discussion

Wandell background discussion

  • To appear: Computational modeling of responses in human visual cortex. BA Wandell, J Winawer, KN Kay.

In Brain Mapping: An Encyclopedic Reference (Edited by Thompson and Friston.) pdf

(Friday) Nov. 14 11 a.m., 560 Evans Hall
Brian Wandell lecture


Nov. 18 Consciousness discussion
Nov. 25

Koch background discussion

  • Koch, C. Project MindScope pdf
  • Tsuchiya, N., & Koch, C. (2008). The relationship between consciousness and attention. The neurology of consciousness: Cognitive neuroscience and neuropathology, 63-78. pdf
  • Klein, S. A. (1993). Will robots see? Chapter in Spatial Vision in Humans and Robots, Cambridge University Press, 184-199. pdf
  • Tononi, G., & Koch, C. (2014). Consciousness: Here, There but Not Everywhere. arXiv preprint arXiv:1405.7089. pdf
  • Scientific American article
  • Scientific American article
Dec. 2, 4-6 p.m., 125 Li Ka Shing
Christof Koch lecture: Unsolved Problems in Vision: Consciousness.

Evening seminar: Koch Discussion


Additional Materials

  • recent special issue of CurrOpinNeuro journal
  • Olshausen BA Olshausen (2013) Perception as an Inference Problem. pdf
  • Olshausen BA (2012) 20 years of learning about vision: Questions answered, questions unanswered, and questions not yet asked. In: 20 Years of Computational Neuroscience (Symposium of the CNS 2010 annual meeting) pdf
  • Kitaoka, A (2014) Color-dependent motion illusions in stationary images and their phenomenal dimorphism. Perception advance online publication pdf
  • O'Regan, J. K., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and brain sciences, 24(05), 939-973.pdf
  • Bruno Olshausen lecture (1 July 2014) 20 Years of Learning About Vision: Questions Answered, Questions Unanswered, and Questions Not Yet Asked video
  • Solari, S. V. H., & Stoner, R. (2011). Cognitive consilience: primate non-primary neuroanatomical circuits underlying cognition. Frontiers in neuroanatomy, 5. pdf
  • Dyson, Freeman. The Case for Blunders. The New York Review of Books, 6 March 2014. pdf
  • Machine-Learning Maestro Michael Jordan on the Delusions of Big Data and Other Huge Engineering Efforts, 20 Oct 2014, Lee Gomes, IEEE Specturm link
  • Yann LeCunn responds to Mike Jordan's Spectrum interview link
  • Kravitz, D. J., Saleem, K. S., Baker, C. I., Ungerleider, L. G., & Mishkin, M. (2013). The ventral visual pathway: an expanded neural framework for the processing of object quality. Trends in cognitive sciences, 17(1), 26-49. pdf
  • Vinyals, O. et al. Show and Tell: A Neural Image Caption Generator. 2014 arXiv.1411.4555v1 pdf
  • Koch, C., & Tononi, G. (2011). A test for consciousness. Scientific American, 304(6), 44-47. [pdf https://www.dropbox.com/s/h2bo3swrjr1g1l1/A_Test_for_Consciousness.pdf?dl=0]