Research
My research interests broadly span CV and ML.
In particular, I am interested in the cryo-EM 3D reconstruction.
I design neural implicit representations as well as pose estimation methods to be used in 3D structure determination in cryo-EM.
|
|
CryoSPIN: Improving Ab-Initio Cryo-EM Reconstruction with Semi-Amortized Pose Inference
Shayan Shekarforoush,
David Lindell,
Marcus Brubaker,
David Fleet
NeurIPS 2024 (+ Oral in ML for Structural Biology workshop)
arXiv /
project page /
code
We introduce cryoSPIN, a new approach to ab-initio cryo-EM 3D reconstruction using semi-amortization to accelerate convergence and multi-head encoder to handle pose uncertainty.
|
|
Dual-Camera Joint Deblurring-Denoising
Shayan Shekarforoush,
Aman Walia,
Marcus Brubaker,
Kosta Derpanis,
Alex Levinshtein
arXiv /
project page
A new method for joint image
deblurring-denoising using a burst of short exposure images
synchronized with a long exposure image, captured using a
dual-camera rig.
Deblurring is guided with flow/motion information estimated from the burst.
|
|
Residual Multiplicative Filter Networks for Multiscale Reconstruction
Shayan Shekarforoush,
David Lindell,
David Fleet,
Marcus Brubaker
NeurIPS 2022
arXiv /
project page /
code
A new coordinate network and training scheme for optimizing multi-scale representation in a coarse-to-fine fashion.
|
|
Physics aware inference for the cryo-EM inverse problem
Geoffrey Woollard,
Shayan Shekarforoush,
Frank Wood,
Marcus Brubaker,
Khanh Dao Duc
NeurIPS Workshop 2022, ML in Structural Biology
paper
Stochastic Variational Inference of pose, defocus and conformational heterogeneity of atomic models while using anisotropic network to model variations in conformations.
|
|
Graph Convolution Based Attention Model for Personalized Disease Prediction
Anees Kazi,
Shayan Shekarforoush,
S.Arvind Krishna,
Hendrik Burwinkel,
Gerome Vivar,
Benedict Wiestler,
Karsten Kortum,
Seyed-Ahmad Ahmadi,
Shadi Albarqouni,
Nassir Navab
MICCAI 2019
Combining LSTM-based attention mechanism and GCNs.
|
|
InceptionGCN: Receptive Field Aware Graph Convolutional Network for Disease Prediction
Anees Kazi,
Shayan Shekarforoush,
S.Arvind Krishna,
Hendrik Burwinkel,
Gerome Vivar,
Karsten Kortuem,
Seyed-Ahmad Ahmadi,
Shadi Albarqouni,
Nassir Navab
IPMI 2019 (Oral Presentation)
arXiv /
code
We define geometric Inception modules capable of capturing intra and inter-graph structural heterogeneity thanks to multiple kernels of different sizes.
|
|
Self-Attention Equipped Graph Convolutions for Disease Prediction
Anees Kazi,
S.Arvind Krishna,
Shayan Shekarforoush,
Karsten Kortuem,
Shadi Albarqouni,
Nassir Navab
ISBI 2019
arXiv
Introducing self-attention layer to GCNs which learns relations between indiviudal demographic data and diseases.
|
|