Deep Feedback Inverse Problem Solver
© 2020, Springer Nature Switzerland AG. We present an efficient, effective, and generic approach towards solving inverse problems. The key idea is to leverage the feedback signal provided by the forward process and learn an iterative update model. Specifically, at each iteration, the neural network...
| Main Authors: | , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Springer International Publishing
2021
|
| Online Access: | https://hdl.handle.net/1721.1/137600 |
| _version_ | 1826214789159845888 |
|---|---|
| author | Ma, WC Wang, S Gu, J Manivasagam, S Torralba, A Urtasun, R |
| author2 | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science |
| author_facet | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Ma, WC Wang, S Gu, J Manivasagam, S Torralba, A Urtasun, R |
| author_sort | Ma, WC |
| collection | MIT |
| description | © 2020, Springer Nature Switzerland AG. We present an efficient, effective, and generic approach towards solving inverse problems. The key idea is to leverage the feedback signal provided by the forward process and learn an iterative update model. Specifically, at each iteration, the neural network takes the feedback as input and outputs an update on current estimation. Our approach does not have any restrictions on the forward process; it does not require any prior knowledge either. Through the feedback information, our model not only can produce accurate estimations that are coherent to the input observation but also is capable of recovering from early incorrect predictions. We verify the performance of our model over a wide range of inverse problems, including 6-DoF pose estimation, illumination estimation, as well as inverse kinematics. Comparing to traditional optimization-based methods, we can achieve comparable or better performance while being two to three orders of magnitude faster. Compared to deep learning-based approaches, our model consistently improves the performance on all metrics. |
| first_indexed | 2024-09-23T16:11:30Z |
| format | Article |
| id | mit-1721.1/137600 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T16:11:30Z |
| publishDate | 2021 |
| publisher | Springer International Publishing |
| record_format | dspace |
| spelling | mit-1721.1/1376002023-01-17T20:31:17Z Deep Feedback Inverse Problem Solver Ma, WC Wang, S Gu, J Manivasagam, S Torralba, A Urtasun, R Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory © 2020, Springer Nature Switzerland AG. We present an efficient, effective, and generic approach towards solving inverse problems. The key idea is to leverage the feedback signal provided by the forward process and learn an iterative update model. Specifically, at each iteration, the neural network takes the feedback as input and outputs an update on current estimation. Our approach does not have any restrictions on the forward process; it does not require any prior knowledge either. Through the feedback information, our model not only can produce accurate estimations that are coherent to the input observation but also is capable of recovering from early incorrect predictions. We verify the performance of our model over a wide range of inverse problems, including 6-DoF pose estimation, illumination estimation, as well as inverse kinematics. Comparing to traditional optimization-based methods, we can achieve comparable or better performance while being two to three orders of magnitude faster. Compared to deep learning-based approaches, our model consistently improves the performance on all metrics. 2021-11-05T19:36:10Z 2021-11-05T19:36:10Z 2020 2021-01-28T14:43:51Z Article http://purl.org/eprint/type/ConferencePaper https://hdl.handle.net/1721.1/137600 Ma, WC, Wang, S, Gu, J, Manivasagam, S, Torralba, A et al. 2020. "Deep Feedback Inverse Problem Solver." Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 12350 LNCS. en 10.1007/978-3-030-58558-7_14 Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf Springer International Publishing MIT web domain |
| spellingShingle | Ma, WC Wang, S Gu, J Manivasagam, S Torralba, A Urtasun, R Deep Feedback Inverse Problem Solver |
| title | Deep Feedback Inverse Problem Solver |
| title_full | Deep Feedback Inverse Problem Solver |
| title_fullStr | Deep Feedback Inverse Problem Solver |
| title_full_unstemmed | Deep Feedback Inverse Problem Solver |
| title_short | Deep Feedback Inverse Problem Solver |
| title_sort | deep feedback inverse problem solver |
| url | https://hdl.handle.net/1721.1/137600 |
| work_keys_str_mv | AT mawc deepfeedbackinverseproblemsolver AT wangs deepfeedbackinverseproblemsolver AT guj deepfeedbackinverseproblemsolver AT manivasagams deepfeedbackinverseproblemsolver AT torralbaa deepfeedbackinverseproblemsolver AT urtasunr deepfeedbackinverseproblemsolver |