Locally orderless tensor networks for classifying two- and three-dimensional medical images
Research output: Contribution to journal › Conference article › Research › peer-review
Standard
Locally orderless tensor networks for classifying two- and three-dimensional medical images. / Selvan, Raghavendra; Ørting, Silas; Dam, Erik B.
In: The Journal of Machine Learning for Biomedical Imaging, Vol. 5, No. SI, 2021, p. 1-21.Research output: Contribution to journal › Conference article › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Locally orderless tensor networks for classifying two- and three-dimensional medical images
AU - Selvan, Raghavendra
AU - Ørting, Silas
AU - Dam, Erik B
N1 - Source code will be available at https://github.com/raghavian/LoTeNet_pytorch/
PY - 2021
Y1 - 2021
N2 - Tensor networks are factorisations of high rank tensors into networks of lower rank tensors and have primarily been used to analyse quantum many-body problems. Tensor networks have seen a recent surge of interest in relation to supervised learning tasks with a focus on image classification. In this work, we improve upon the matrix product state (MPS) tensor networks that can operate on one-dimensional vectors to be useful for working with 2D and 3D medical images. We treat small image regions as orderless, squeeze their spatial information into feature dimensions and then perform MPS operations on these locally orderless regions. These local representations are then aggregated in a hierarchical manner to retain global structure. The proposed locally orderless tensor network (LoTeNet) is compared with relevant methods on three datasets. The architecture of LoTeNet is fixed in all experiments and we show it requires lesser computational resources to attain performance on par or superior to the compared methods.
AB - Tensor networks are factorisations of high rank tensors into networks of lower rank tensors and have primarily been used to analyse quantum many-body problems. Tensor networks have seen a recent surge of interest in relation to supervised learning tasks with a focus on image classification. In this work, we improve upon the matrix product state (MPS) tensor networks that can operate on one-dimensional vectors to be useful for working with 2D and 3D medical images. We treat small image regions as orderless, squeeze their spatial information into feature dimensions and then perform MPS operations on these locally orderless regions. These local representations are then aggregated in a hierarchical manner to retain global structure. The proposed locally orderless tensor network (LoTeNet) is compared with relevant methods on three datasets. The architecture of LoTeNet is fixed in all experiments and we show it requires lesser computational resources to attain performance on par or superior to the compared methods.
KW - cs.CV
KW - cs.LG
KW - stat.ML
M3 - Conference article
VL - 5
SP - 1
EP - 21
JO - The Journal of Machine Learning for Biomedical Imaging
JF - The Journal of Machine Learning for Biomedical Imaging
SN - 2766-905X
IS - SI
T2 - MIDL 2020 : International Conference on Medical Imaging with Deep Learning
Y2 - 6 July 2020 through 8 July 2020
ER -
ID: 249297561