Research & Publications

Two questions have guided my scientific research:

1. How does network structure influence computational function in neural systems?
2. How can complex systems use experience to improve performance?

The papers below reflect ongoing progress on these issues from various points of view: theoretical analysis of the relationship between structure and function [3, 5], engineering of artificial systems that learn from data [4, 5, 7, 8], and “connectomics” — reconstruction of biological neural networks for insight into real nervous systems [9, 10, 11]. Undergraduate work veered into nonlinear dimensionality reduction [2] and statistical machine translation [1].

Most of these publications can also be accessed via Google Scholar.


[17] Morphological Error Detection in 3D Segmentations
Rolnick, Meirovitch, Parag, Pfister, Jain, Lichtman, Boyden, Shavit.
arXiv pre-print, 2017

[16] Superhuman Accuracy on the SNEMI3D Connectomics Challenge
Lee, Zung, Li, Jain, and Seung.
arXiv pre-print, 2017

[15] Flood-filling Networks.
Januszewski, Maitin-Shepard, Li, Kornfeld, Denk, and Jain.
arXiv pre-print, 2016

[14] Combinatorial Energy Learning for Image Segmentation.
Maitin-Shepard, Jain, Januszewski, Li, and Abbeel.
Advances in Neural Information Processing Systems 29 [NIPS 2016]

[13] Deep and Wide Multiscale Recursive Networks for Robust Image Labeling.
Huang and Jain.
International Conference on Learning Representations [ICLR 2014; Conference Track]

[12] Learned versus Hand-Designed Feature Representations for 3d Agglomeration.
Bogovic, Huang, and Jain.
International Conference on Learning Representations [ICLR 2014; Conference Track & Oral]

[11] Connectomic reconstruction of the inner plexiform layer in the mouse retina.
Helmstaedter, Briggman, Turaga, Jain, Seung, and Denk.
Nature, 2013

[10] Learning to Agglomerate Superpixel Hierarchies.
Jain, Turaga, Briggman, Helmstaedter, Denk, and Seung.
Advances in Neural Information Processing Systems 24 [NIPS 2011]

[9] Machines that learn to segment images: a crucial technology for connectomics.
Jain, Seung, and Turaga.
Current Opinion in Neurobiology, 2010

[8] Boundary Learning by Optimization with Topological Constraints.
Jain, Bollmann, Richardson, Berger, Helmstaedter, Briggman, Denk, Bowden, Mendenhall, Abraham, Harris, Kasthuri, Hayworth, Schalek, Tapia, Lichtman, and Seung.
IEEE Conference on Computer Vision and Pattern Recognition [CVPR 2010]
[paper] [supplementary: specific methods]

[7] Convolutional networks can learn to generate affinity graphs for image segmentation.
Turaga, Murray, Jain, Roth, Helmstaedter, Briggman, Denk, and Seung.
Neural Computation, 2010

[6] Machine Learning of Image Analysis with Convolutional Networks and Topological Constraints.
PhD Dissertation, Massachusetts Institute of Technology, 2009

[5] Natural Image Denoising with Convolutional Networks.
Jain and Seung.
Advances in Neural Information Processing Systems 21 [NIPS 2008]
[paper] [sample implementation: archive, readme]

[4] Supervised Learning of Image Restoration with Convolutional Networks.
Jain, Murray, Roth, Turaga, Zhigulin, Briggman, Helmstaedter, Denk, and Seung.
International Conference on Computer Vision [ICCV 2007]
[paper] [supplementary: specific methods]

[3] Representing Part-Whole Relations in Recurrent Neural Networks.
Jain, Zhigulin, and Seung.
Advances in Neural Information Processing Systems 18 [NIPS 2005]

[2] Exploratory analysis and visualization of speech and music by locally linear embedding.
Jain and Saul.
International Conf. on Speech, Acoustics, & Signal Processing [ICASSP 2004]

[1] A Smorgasbord of Features for Statistical Machine Translation.
Och, Gildea, Khudanpur, Sarkar, Yamada, Fraser, Kumar, Shen, Smith, Eng, Jain, Jin, and Radev.
Human Language Technology and 5th Meeting of the NAACL [HLT-NAACL 2004]