2018 IEEE International Conference on Robotics and Automation (ICRA) 2018
DOI: 10.1109/icra.2018.8460828
|View full text |Cite
|
Sign up to set email alerts
|

What is (Missing or Wrong) in the Scene? A Hybrid Deep Boltzmann Machine for Contextualized Scene Modeling

Abstract: Scene models allow robots to reason about what is in the scene, what else should be in it, and what should not be in it. In this paper, we propose a hybrid Boltzmann Machine (BM) for scene modeling where relations between objects are integrated. To be able to do that, we extend BM to include tri-way edges between visible (object) nodes and make the network to share the relations across different objects. We evaluate our method against several baseline models (Deep Boltzmann Machines, and Restricted Boltzmann M… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

1
4
0

Year Published

2019
2019
2022
2022

Publication Types

Select...
4
3
1

Relationship

2
6

Authors

Journals

citations
Cited by 8 publications
(5 citation statements)
references
References 29 publications
1
4
0
Order By: Relevance
“…In this case, the added object can dominate the scene context and the COSMO removes an object that belongs to the groundtruth sample, and it corrupts original input data. The same problem has been observed in our previous work [47].…”
Section: Discussion Of Resultssupporting
confidence: 87%
See 1 more Smart Citation
“…In this case, the added object can dominate the scene context and the COSMO removes an object that belongs to the groundtruth sample, and it corrupts original input data. The same problem has been observed in our previous work [47].…”
Section: Discussion Of Resultssupporting
confidence: 87%
“…Numerical comparison of results with our previous work [47] is not directly feasible since (i) different measures are used in Task 1 and Task 2, (ii) the current dataset is more difficult and (iii) the affordance representation is added to the model. In this work, the training set is richer in scenes with more objects and relations compared to the dataset in our previous work.…”
Section: Quantitative Comparison With Our Previous Workmentioning
confidence: 99%
“…A BM is entirely related between and within the layers, while an RBM excludes the lateral relations in hidden and apparent layers. The random variables represented by secret units are, therefore, not conditional in terms of the circumstances of the exposed groups, and vice versa [26].…”
Section: Boltzmann Machine (Bm)mentioning
confidence: 99%
“…A BM is fully connected between and within the layers, while in a RBM, the lateral connections in the hidden and visible layers are eliminated. Therefore, the random variables which are encoded by hidden units are not conditionally dependent taking under consideration the states of the visible units, and the other way around [26]. Ni GAO et al suggested an approach which has been based on the multilayer DBN for the DoS attacks detection.…”
Section: Fig 4 Comparison Of Boltzmann Machines Restricted Boltzmamentioning
confidence: 99%