Model-based Phase Retrieval with Deep Denoiser Prior

2020-06-26
Işıl, Çağatay
Öktem, Sevinç Figen
Show full item record
116
views
0
downloads
We develop a novel phase-retrieval algorithm with deep denoiser prior. The approach incorporates learning-based prior to the hybrid input-output method through plug- and-play regularization. Results demonstrate the state-of-the-art performance of our approach and its computational efficiency.
https://hdl.handle.net/11511/80121
https://www.osapublishing.org/abstract.cfm?uri=COSI-2020-CF2C.5
Department of Electrical and Electronics Engineering, Conference / Seminar

Suggestions

Deep Learning-Based Hybrid Approach for Phase Retrieval
IŞIL, ÇAĞATAY; Öktem, Sevinç Figen; KOÇ, AYKUT (2019-06-24)
We develop a phase retrieval algorithm that utilizes the hybrid-input-output (HIO) algorithm with a deep neural network (DNN). The DNN architecture, which is trained to remove the artifacts of HIO, is used iteratively with HIO to improve the reconstructions. The results demonstrate the effectiveness of the approach with little additional cost.
Comparison of tissue disease specific integrated networks using directed graphlet signatures
SÖNMEZ, ARZU BURÇAK; Can, Tolga (2016-10-05)
We present a novel framework for counting small sub-graph patterns in integrated genome-scale networks. An integrated network was built using the physical, regulatory, and metabolic interactions between H. sapiens proteins from the Pathway Commons database. The network was filtered for tissue/disease specific proteins by using a large-scale human transcriptional profiling study, resulting in several tissue and disease specific sub-networks. In this study, we apply and extend the idea of graphlet counting in...
Efficient parallelization of multilevel fast multipole algorithm
Ergül, Özgür Salih (null; 2006-11-10)
We report our efforts for the solution of large electromagnetics problems accurately and efficiently with the parallel multilevel fast multipole algorithm. We carefully investigate different stages of the parallelization and identify the bottlenecks to develop new strategies. The required modifications are implemented in order to increase the efficiency of the solutions of scattering problems involving various geometries.
Optimisation of dragline inspection intervals with time-counter algorithm
Gölbaşı, Onur; Demirel, Nuray (Informa UK Limited, 2017-01-01)
This research study proposed a novel algorithm, called time-counter, to optimise inspection intervals of production systems. The developed algorithm was applied to two active draglines. The algorithm evaluated random uptime/downtime characteristics of the draglines and the planned dragline halts in the mine. Physical and random non-physical costs were included and annual maintenance cost minimisation was achieved. Dragline-1 and Dragline-2 inspection intervals were optimised as 232 and 184 h and their maint...
Linear Planning Logic: An Efficient Language and Theorem Prover for Robotic Task Planning
Kortik, Sitar; Saranlı, Uluç (2014-06-07)
In this paper, we introduce a novel logic language and theorem prover for robotic task planning. Our language, which we call Linear Planning Logic (LPL), is a fragment of linear logic whose resource-conscious semantics are well suited for reasoning with dynamic state, while its structure admits efficient theorem provers for automatic plan construction. LPL can be considered as an extension of Linear Hereditary Harrop Formulas (LHHF), whose careful design allows the minimization of nondeterminism in proof se...
Citation Formats
Ç. Işıl and S. F. Öktem, “Model-based Phase Retrieval with Deep Denoiser Prior,” 2020, Accessed: 00, 2021. [Online]. Available: https://hdl.handle.net/11511/80121.
METU IIS - Integrated Information Service open@metu.edu.tr