+publisher:"University of Texas – Austin" +contributor:("Niekum, Scott")

University of Texas – Austin

1. -7411-0398. Data efficient reinforcement learning with off-policy and simulated data.

Degree: PhD, Computer Science, 2019, University of Texas – Austin

URL: http://dx.doi.org/10.26153/tsw/7716

► Learning from interaction with the environment – trying untested actions, observing successes and failures, and tying effects back to causes – is one of the… (more)

▼ Learning from interaction with the environment – trying untested actions, observing successes and failures, and tying effects back to causes – is one of the first capabilities we think of when considering autonomous agents. Reinforcement learning (RL) is the area of artificial intelligence research that has the goal of allowing autonomous agents to learn in this way. Despite much recent success, many modern reinforcement learning algorithms are still limited by the requirement of large amounts of experience before useful skills are learned. Two possible approaches to improving data efficiency are to allow algorithms to make better use of past experience collected with past behaviors (known as off-policy data) and to allow algorithms to make better use of simulated data sources. This dissertation investigates the use of such auxiliary data by answering the question, "How can a reinforcement learning agent leverage off-policy and simulated data to evaluate and improve upon the expected performance of a policy?" This dissertation first considers how to directly use off-policy data in reinforcement learning through importance sampling. When used in reinforcement learning, importance sampling is limited by high variance that leads to inaccurate estimates. This dissertation addresses this limitation in two ways. First, this dissertation introduces the behavior policy gradient algorithm that adapts the data collection policy towards a policy that generates data that leads to low variance importance sampling evaluation of a fixed policy. Second, this dissertation introduces the family of regression importance sampling estimators which improve the weighting of already collected off-policy data so as to lower the variance of importance sampling evaluation of a fixed policy. In addition to evaluation of a fixed policy, we apply the behavior policy gradient algorithm and regression importance sampling to batch policy gradient policy improvement. In the case of regression importance sampling, this application leads to the introduction of the sampling error corrected policy gradient estimator that improves the data efficiency of batch policy gradient algorithms. Towards the goal of learning from simulated experience, this dissertation introduces an algorithm – the grounded action transformation algorithm – that takes small amounts of real world data and modifies the simulator such that skills learned in simulation are more likely to carry over to the real world. Key to this approach is the idea of local simulator modification – the simulator is automatically altered to better model the real world for actions the data collection policy would take in states the data collection policy would visit. Local modification necessitates an iterative approach: the simulator is modified, the policy improved, and then more data is collected for further modification. Finally, in addition to examining them each independently, this dissertation also considers the possibility of combining the use of simulated data with importance sampled… Advisors/Committee Members: Stone, Peter, 1971- (advisor), Niekum, Scott (committee member), Krähenbühl, Philipp (committee member), Sutton, Richard (committee member).

Subjects/Keywords: Artificial intelligence; Reinforcement learning; Robotics; Off-policy; Sim-to-real

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

-7411-0398. (2019). Data efficient reinforcement learning with off-policy and simulated data. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://dx.doi.org/10.26153/tsw/7716

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Chicago Manual of Style (16^th Edition):

-7411-0398. “Data efficient reinforcement learning with off-policy and simulated data.” 2019. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://dx.doi.org/10.26153/tsw/7716.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

MLA Handbook (7^th Edition):

-7411-0398. “Data efficient reinforcement learning with off-policy and simulated data.” 2019. Web. 05 Jul 2020.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Vancouver:

-7411-0398. Data efficient reinforcement learning with off-policy and simulated data. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2019. [cited 2020 Jul 05]. Available from: http://dx.doi.org/10.26153/tsw/7716.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Council of Science Editors:

-7411-0398. Data efficient reinforcement learning with off-policy and simulated data. [Doctoral Dissertation]. University of Texas – Austin; 2019. Available from: http://dx.doi.org/10.26153/tsw/7716

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

2. -6888-3095. Embodied learning for visual recognition.

Degree: PhD, Electrical and Computer Engineering, 2017, University of Texas – Austin

URL: http://hdl.handle.net/2152/63489

► The field of visual recognition in recent years has come to rely on large expensively curated and manually labeled "bags of disembodied images". In the… (more)

Subjects/Keywords: Computer vision; Unsupervised learning; Embodied learning

10 more images…

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

-6888-3095. (2017). Embodied learning for visual recognition. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/63489

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Chicago Manual of Style (16^th Edition):

-6888-3095. “Embodied learning for visual recognition.” 2017. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/63489.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

MLA Handbook (7^th Edition):

-6888-3095. “Embodied learning for visual recognition.” 2017. Web. 05 Jul 2020.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Vancouver:

-6888-3095. Embodied learning for visual recognition. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2017. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/63489.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Council of Science Editors:

-6888-3095. Embodied learning for visual recognition. [Doctoral Dissertation]. University of Texas – Austin; 2017. Available from: http://hdl.handle.net/2152/63489

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

3. Liebman, Elad. Sequential decision making in artificial musical intelligence.

Degree: PhD, Computer Science, 2019, University of Texas – Austin

URL: http://dx.doi.org/10.26153/tsw/2858

► Over the past 60 years, artificial intelligence has grown from a largely academic field of research to a ubiquitous array of tools and approaches used… (more)

▼ Over the past 60 years, artificial intelligence has grown from a largely academic field of research to a ubiquitous array of tools and approaches used in everyday technology. Despite its many recent successes and growing prevalence, certain meaningful facets of computational intelligence have not been as thoroughly explored. Such additional facets cover a wide array of complex mental tasks which humans carry out easily, yet are difficult for computers to mimic. A prime example of a domain in which human intelligence thrives, but machine understanding is still fairly limited, is music. Over the last decade, many researchers have applied computational tools to carry out tasks such as genre identification, music summarization, music database querying, and melodic segmentation. While these are all useful algorithmic solutions, we are still a long way from constructing complete music agents, able to mimic (at least partially) the complexity with which humans approach music. One key aspect which hasn't been sufficiently studied is that of sequential decision making in musical intelligence. This thesis strives to answer the following question: Can a sequential decision making perspective guide us in the creation of better music agents, and social agents in general? And if so, how? More specifically, this thesis focuses on two aspects of musical intelligence: music recommendation and human-agent (and more generally agent-agent) interaction in the context of music. The key contributions of this thesis are the design of better music playlist recommendation algorithms; the design of algorithms for tracking user preferences over time; new approaches for modeling people's behavior in situations that involve music; and the design of agents capable of meaningful interaction with humans and other agents in a setting where music plays a roll (either directly or indirectly). Though motivated primarily by music-related tasks, and focusing largely on people's musical preferences, this thesis also establishes that insights from music-specific case studies can also be applicable in other concrete social domains, such as different types of content recommendation. Showing the generality of insights from musical data in other contexts serves as evidence for the utility of music domains as testbeds for the development of general artificial intelligence techniques. Ultimately, this thesis demonstrates the overall usefulness of taking a sequential decision making approach in settings previously unexplored from this perspective Advisors/Committee Members: Stone, Peter, 1971- (advisor), Dannenberg, Roger (committee member), Grauman, Kristen (committee member), Niekum, Scott (committee member), Saar-Tsechansky, Maytal (committee member).

Subjects/Keywords: Music informatics; Reinforcement learning; Artificial intelligence; Sequential decision-making

10 more images…

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Liebman, E. (2019). Sequential decision making in artificial musical intelligence. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://dx.doi.org/10.26153/tsw/2858

Chicago Manual of Style (16^th Edition):

Liebman, Elad. “Sequential decision making in artificial musical intelligence.” 2019. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://dx.doi.org/10.26153/tsw/2858.

MLA Handbook (7^th Edition):

Liebman, Elad. “Sequential decision making in artificial musical intelligence.” 2019. Web. 05 Jul 2020.

Vancouver:

Liebman E. Sequential decision making in artificial musical intelligence. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2019. [cited 2020 Jul 05]. Available from: http://dx.doi.org/10.26153/tsw/2858.

Council of Science Editors:

Liebman E. Sequential decision making in artificial musical intelligence. [Doctoral Dissertation]. University of Texas – Austin; 2019. Available from: http://dx.doi.org/10.26153/tsw/2858

4. Mahjourian, Reza. Hierarchical policy design for sample-efficient learning of robot table tennis through self-play.

Degree: PhD, Computer Science, 2019, University of Texas – Austin

URL: http://hdl.handle.net/2152/72812

► Training robots with physical bodies requires developing new methods and action representations that allow the learning agents to explore the space of policies efficiently. This… (more)

▼ Training robots with physical bodies requires developing new methods and action representations that allow the learning agents to explore the space of policies efficiently. This work studies sample-efficient learning of complex policies in the context of robot table tennis. It incorporates learning into a hierarchical control framework using a model-free strategy layer (which requires complex reasoning about opponents that is difficult to do in a model-based way), model-based prediction of external objects (which are difficult to control directly with analytic control methods, but governed by learnable and relatively simple laws of physics), and analytic controllers for the robot itself. Human demonstrations are used to train dynamics models, which together with the analytic controller allow any robot that is physically capable to play table tennis without training episodes. Using only about 7000 demonstrated trajectories, a striking policy can hit ball targets with about 20 cm error. Self-play is used to train cooperative and adversarial strategies on top of model-based striking skills trained from human demonstrations. After only about 24000 strikes in self-play the agent learns to best exploit the human dynamics models for longer cooperative games. Further experiments demonstrate that more flexible variants of the policy can discover new strikes not demonstrated by humans and achieve higher performance at the expense of lower sample-efficiency. Experiments are carried out in a virtual reality environment using sensory observations that are obtainable in the real world. The high sample-efficiency demonstrated in the evaluations show that the proposed method is suitable for learning directly on physical robots without transfer of models or policies from simulation. Advisors/Committee Members: Miikkulainen, Risto (advisor), Levine, Sergey (committee member), Sentis, Luis (committee member), Niekum, Scott (committee member), Mok, Aloysius (committee member).

Subjects/Keywords: Robotics; Table tennis; Self-play; Reinforcement learning; Hierarchical policy

10 more images…

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Mahjourian, R. (2019). Hierarchical policy design for sample-efficient learning of robot table tennis through self-play. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/72812

Chicago Manual of Style (16^th Edition):

Mahjourian, Reza. “Hierarchical policy design for sample-efficient learning of robot table tennis through self-play.” 2019. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/72812.

MLA Handbook (7^th Edition):

Mahjourian, Reza. “Hierarchical policy design for sample-efficient learning of robot table tennis through self-play.” 2019. Web. 05 Jul 2020.

Vancouver:

Mahjourian R. Hierarchical policy design for sample-efficient learning of robot table tennis through self-play. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2019. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/72812.

Council of Science Editors:

Mahjourian R. Hierarchical policy design for sample-efficient learning of robot table tennis through self-play. [Doctoral Dissertation]. University of Texas – Austin; 2019. Available from: http://hdl.handle.net/2152/72812

University of Texas – Austin

5. -4766-7868. Self-learning control of automated drilling operations.

Degree: PhD, Mechanical Engineering, 2018, University of Texas – Austin

URL: http://hdl.handle.net/2152/65829

► In recent years, drilling automation has sparked significant interest in both the upstream oil and gas industry and the drilling research community. Automation of various… (more)

▼ In recent years, drilling automation has sparked significant interest in both the upstream oil and gas industry and the drilling research community. Automation of various drilling tasks can potentially allow for higher operational efficiency, increased consistency, and reduced risk of trouble events. However, wide adoption of drilling automation has been slow. This can be primarily attributed to the complex nature of drilling, and the high variability in well types and rig specifications that prevent the deployment of off-the-shelf automation solutions. Such complexities justify the need for an automation system that can self-learn by interacting with the drilling environment to reduce uncertainty. The aim of this dissertation is to determine how a drilling automation system can learn from the environment and utilize this learning to control drilling tasks optimally. To provide an answer, the importance of learning, as well as its limitations in dealing with challenges such as insufficient training data, are explored. A self-learning control system is presented that addresses the aforementioned research question in the context of optimization, control, and event detection. By adopting an action-driven learning approach, the control system can learn the parameters that describe system dynamics. An action-driven approach is shown to also enable the learning of the relationship between control actions and user-defined performance metrics. The resulting knowledge of this learning process enables the system to make and execute optimal decisions without relying on simplifying assumptions that are often made in the drilling literature. Detection of trouble drilling events is explored, and methods for reduction of false/missed alarms are presented to minimize false interruptions of the drilling control system. The subcomponents of the self-learning control system are validated using simulated and actual field data from drilling operations to ascertain the effectiveness of the proposed methods. Advisors/Committee Members: Oort, Eric van (advisor), Fernandez, Benito R. (advisor), Chen, Dongmei (committee member), Barr, Ronald E. (committee member), Niekum, Scott (committee member).

Subjects/Keywords: Automated drilling; Drilling optimization; Self-learning control

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

-4766-7868. (2018). Self-learning control of automated drilling operations. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/65829

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Chicago Manual of Style (16^th Edition):

-4766-7868. “Self-learning control of automated drilling operations.” 2018. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/65829.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

MLA Handbook (7^th Edition):

-4766-7868. “Self-learning control of automated drilling operations.” 2018. Web. 05 Jul 2020.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Vancouver:

-4766-7868. Self-learning control of automated drilling operations. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2018. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/65829.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Council of Science Editors:

-4766-7868. Self-learning control of automated drilling operations. [Doctoral Dissertation]. University of Texas – Austin; 2018. Available from: http://hdl.handle.net/2152/65829

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

University of Texas – Austin

6. -6763-2625. Multilayered skill learning and movement coordination for autonomous robotic agents.

Degree: PhD, Computer Science, 2017, University of Texas – Austin

URL: http://hdl.handle.net/2152/62889

► With advances in technology expanding the capabilities of robots, while at the same time making robots cheaper to manufacture, robots are rapidly becoming more prevalent… (more)

▼ With advances in technology expanding the capabilities of robots, while at the same time making robots cheaper to manufacture, robots are rapidly becoming more prevalent in both industrial and domestic settings. An increase in the number of robots, and the likely subsequent decrease in the ratio of people currently trained to directly control the robots, engenders a need for robots to be able to act autonomously. Larger numbers of robots present together provide new challenges and opportunities for developing complex autonomous robot behaviors capable of multirobot collaboration and coordination. The focus of this thesis is twofold. The first part explores applying machine learning techniques to teach simulated humanoid robots skills such as how to move or walk and manipulate objects in their environment. Learning is performed using reinforcement learning policy search methods, and layered learning methodologies are employed during the learning process in which multiple lower level skills are incrementally learned and combined with each other to develop richer higher level skills. By incrementally learning skills in layers such that new skills are learned in the presence of previously learned skills, as opposed to individually in isolation, we ensure that the learned skills will work well together and can be combined to perform complex behaviors (e.g. playing soccer). The second part of the thesis centers on developing algorithms to coordinate the movement and efforts of multiple robots working together to quickly complete tasks. These algorithms prioritize minimizing the makespan, or time for all robots to complete a task, while also attempting to avoid interference and collisions among the robots. An underlying objective of this research is to develop techniques and methodologies that allow autonomous robots to robustly interact with their environment (through skill learning) and with each other (through movement coordination) in order to perform tasks and accomplish goals asked of them. The work in this thesis is implemented and evaluated in the RoboCup 3D simulation soccer domain, and has been a key component of the UT Austin Villa team winning the RoboCup 3D simulation league world championship six out of the past seven years. Advisors/Committee Members: Stone, Peter, 1971- (advisor), Ballard, Dana (committee member), Egerstedt, Magnus (committee member), Miikkulainen, Risto (committee member), Niekum, Scott (committee member).

Subjects/Keywords: Overlapping layered learning; Role assignment; Reinforcement learning; Robotics; Robot soccer

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

-6763-2625. (2017). Multilayered skill learning and movement coordination for autonomous robotic agents. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/62889

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Chicago Manual of Style (16^th Edition):

-6763-2625. “Multilayered skill learning and movement coordination for autonomous robotic agents.” 2017. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/62889.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

MLA Handbook (7^th Edition):

-6763-2625. “Multilayered skill learning and movement coordination for autonomous robotic agents.” 2017. Web. 05 Jul 2020.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Vancouver:

-6763-2625. Multilayered skill learning and movement coordination for autonomous robotic agents. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2017. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/62889.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Council of Science Editors:

-6763-2625. Multilayered skill learning and movement coordination for autonomous robotic agents. [Doctoral Dissertation]. University of Texas – Austin; 2017. Available from: http://hdl.handle.net/2152/62889

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

7. Jain, Suyog Dutt. Human machine collaboration for foreground segmentation in images and videos.

Degree: PhD, Computer Science, 2018, University of Texas – Austin

URL: http://hdl.handle.net/2152/63453

► Foreground segmentation is defined as the problem of generating pixel level foreground masks for all the objects in a given image or video. Accurate foreground… (more)

▼ Foreground segmentation is defined as the problem of generating pixel level foreground masks for all the objects in a given image or video. Accurate foreground segmentations in images and videos have several potential applications such as improving search, training richer object detectors, image synthesis and re-targeting, scene and activity understanding, video summarization, and post-production video editing. One effective way to solve this problem is human-machine collaboration. The main idea is to let humans guide the segmentation process through some partial supervision. As humans, we are extremely good at perception and can easily identify the foreground regions. Computers, on the other hand, lack this capability, but are extremely good at continuously processing large volumes of data at the lowest level of detail with great efficiency. Bringing these complementary strengths together can lead to systems which are accurate and cost-effective at the same time. However, in any such human-machine collaboration system, cost effectiveness and higher accuracy are competing goals. While more involvement from humans can certainly lead to higher accuracy, it also leads to increased cost both in terms of time and money. On the other hand, relying more on machines is cost-effective, but algorithms are still nowhere near human-level performance. Balancing this cost versus accuracy trade-off holds the key behind success for such a hybrid system. In this thesis, I develop foreground segmentation algorithms which effectively and efficiently make use of human guidance for accurately segmenting foreground objects in images and videos. The algorithms developed in this thesis actively reason about the best modalities or interactions through which a user can provide guidance to the system for generating accurate segmentations. At the same time, these algorithms are also capable of prioritizing human guidance on instances where it is most needed. Finally, when structural similarity exists within data (e.g., adjacent frames in a video or similar images in a collection), the algorithms developed in this thesis are capable of propagating information from instances which have received human guidance to the ones which did not. Together, these characteristics result in a substantial savings in human annotation cost while generating high quality foreground segmentations in images and videos. In this thesis, I consider three categories of segmentation problems all of which can greatly benefit from human-machine collaboration. First, I consider the problem of interactive image segmentation. In traditional interactive methods a human annotator provides a coarse spatial annotation (e.g., bounding box or freehand outlines) around the object of interest to obtain a segmentation. The mode of manual annotation used affects both its accuracy and ease-of-use. Whereas existing methods assume a fixed form of input no matter the image, in this thesis I propose a data-driven algorithm which learns whether an interactive segmentation method will… Advisors/Committee Members: Grauman, Kristen Lorraine, 1979- (advisor), Mooney, Raymond (committee member), Corso, Jason (committee member), Niekum, Scott (committee member), Vouga, Paul Etienne (committee member).

Subjects/Keywords: Computer vision; Crowdsourcing; Human machine collaboration; Image and video segmentation; Image segmentation; Video segmentation; Foreground segmentation; Object segmentation

10 more images…

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Jain, S. D. (2018). Human machine collaboration for foreground segmentation in images and videos. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/63453

Chicago Manual of Style (16^th Edition):

Jain, Suyog Dutt. “Human machine collaboration for foreground segmentation in images and videos.” 2018. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/63453.

MLA Handbook (7^th Edition):

Jain, Suyog Dutt. “Human machine collaboration for foreground segmentation in images and videos.” 2018. Web. 05 Jul 2020.

Vancouver:

Jain SD. Human machine collaboration for foreground segmentation in images and videos. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2018. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/63453.

Council of Science Editors:

Jain SD. Human machine collaboration for foreground segmentation in images and videos. [Doctoral Dissertation]. University of Texas – Austin; 2018. Available from: http://hdl.handle.net/2152/63453

University of Texas – Austin

8. -2711-6738. Learning for 360° video compression, recognition, and display.

Degree: PhD, Computer Science, 2019, University of Texas – Austin

URL: http://dx.doi.org/10.26153/tsw/5848

► 360° cameras are a core building block of the Virtual Reality (VR) and Augmented Reality (AR) technology that bridges the real and digital worlds. It… (more)

▼ 360° cameras are a core building block of the Virtual Reality (VR) and Augmented Reality (AR) technology that bridges the real and digital worlds. It allows us to build virtual environments for VR/AR applications from the real world easily by capturing the entire visual world surrounding the camera simultaneously. With the rapid growth of VR/AR technology, the availability and popularity of 360° cameras are also growing faster than ever. People now create, share, and watch 360° content in their everyday life, and the amount of 360° content is increasing rapidly. While 360° cameras offer tremendous new possibilities in various domains, they also introduce new technical challenges. These challenges span over the entire 360° video production pipeline, ranging all the way from video capturing to high level applications. For example, the 360° field-of-view makes it difficult to display the content to users, and the distortion in the planar projection degrades the performance of both the compression and visual recognition algorithms. Many of the challenges remain unsolved or even unexplored, which prohibits people from exploiting the full potential of the new media. This leads to a dire need for a more mature 360° video production pipeline like those for traditional media. To this end, my thesis targets three fundamental challenges in 360° production—video compression, visual recognition, and 360° video display. Because a proper compressed format is the foundation of all video technologies and applications, I begin with improving 360° video compression. It has been shown that existing video codecs do not perform well on 360° video, and 360° video compression standard is under rapid development. Complementary to the ongoing progress in 360° video compression standards, I propose to exploit the orientation of the 360° video projection for a better compression rate. The method explores a new dimension in video compression and is compatible with existing compression technologies. It reduces video sizes to allow easy storage and transmission of 360° videos. Besides being able to collect and distribute 360° content, another prerequisite for building advanced applications on top of the new media is the ability to analyze the visual content. Therefore, I next study visual recognition on 360° imagery. I propose a general approach that transfers an existing Convolutional Neural Network (CNN) trained on perspective images to 360° imagery. It allows us to transfer knowledge from perspective images to 360° images, including both the network architecture and training data. Compared with existing strategies for applying existing CNN models on 360° data, the method sacrifices neither accuracy nor efficiency and does not need any additional annotation effort. The method allows us to perform visual recognition on the new format given an existing CNN model with zero manual labor. After building the basis for 360° video applications, I finally tackle one of the most important applications of 360° video: displaying the video content to users.… Advisors/Committee Members: Grauman, Kristen Lorraine, 1979- (advisor), Snavely, Noah (committee member), Huang, Qixing (committee member), Niekum, Scott (committee member).

Subjects/Keywords: 360° video; Omnidirectional media; Video analysis

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

-2711-6738. (2019). Learning for 360° video compression, recognition, and display. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://dx.doi.org/10.26153/tsw/5848

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Chicago Manual of Style (16^th Edition):

-2711-6738. “Learning for 360° video compression, recognition, and display.” 2019. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://dx.doi.org/10.26153/tsw/5848.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

MLA Handbook (7^th Edition):

-2711-6738. “Learning for 360° video compression, recognition, and display.” 2019. Web. 05 Jul 2020.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Vancouver:

-2711-6738. Learning for 360° video compression, recognition, and display. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2019. [cited 2020 Jul 05]. Available from: http://dx.doi.org/10.26153/tsw/5848.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Council of Science Editors:

-2711-6738. Learning for 360° video compression, recognition, and display. [Doctoral Dissertation]. University of Texas – Austin; 2019. Available from: http://dx.doi.org/10.26153/tsw/5848

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

University of Texas – Austin

9. Xiong, Bo. Learning to compose photos and videos from passive cameras.

Degree: PhD, Computer Science, 2019, University of Texas – Austin

URL: http://dx.doi.org/10.26153/tsw/5847

► Photo and video overload is well-known to most computer users. With cameras on mobile devices, it is all too easy to snap images and videos… (more)

▼ Photo and video overload is well-known to most computer users. With cameras on mobile devices, it is all too easy to snap images and videos spontaneously, yet it remains much less easy to organize or search through that content later. With increasingly portable wearable and 360° computing platforms, the overload problem is only intensifying. Wearable and 360° cameras passively record everything they observe, unlike traditional cameras that require active human attention to capture images or videos. In my thesis, I explore the idea of automatically composing photos and videos from unedited videos captured by "passive" cameras. Passive cameras (e.g., wearable cameras, 360° cameras) offer a more relaxing experience to record our visual world but they do not always capture frames that look like intentional human-taken photos. In wearable cameras, many frames will be blurry, contain poorly composed shots, and/or simply have uninteresting content. In 360° cameras, a single omni-directional image captures the entire visual world, and the photographer's intention and attention in that moment are unknown. To this end, I consider the following problems in the context of passive cameras: 1) what visual data to capture and store, 2) how to identify foreground objects, and 3) how to enhance the viewing experience. First, I explore the problem of finding the best moments in unedited videos. Not everything observed in a wearable camera's video stream is worthy of being captured and stored. People can easily distinguish well-composed moments from accidental shots from a wearable camera. This prompts the question: can a vision system predict the best moments in unedited video? I first study how to find the best moments in terms of short video clips. My key insight is that video segments from shorter user-generated videos are more likely to be highlights than those from longer videos, since users tend to be more selective about the content when capturing shorter videos. Leveraging this insight, I introduce a novel ranking framework to learn video highlight detection from unlabeled videos. Next, I show how to predict snap points in unedited video – that is, those frames that look like intentionally taken photos. I propose a framework to detect snap points that requires no human annotations. The main idea is to construct a generative model of what human-taken photos look like by sampling images posted on the Web. Snapshots that people upload to share publicly online may vary vastly in their content, yet all share the key facet that they were intentional snap point moments. This makes them an ideal source of positive exemplars for our target learning problem. In both settings, despite learning without any explicit labels, my proposed models outperform discriminative baselines trained with labeled data. Next, I introduce a novel approach to automatically segment foreground objects in images and videos. Identifying key objects is an important intermediate step for automatic photo composition. It is also a prerequisite in graphics… Advisors/Committee Members: Grauman, Kristen Lorraine, 1979- (advisor), Hays, James (committee member), Huang, Qixing (committee member), Niekum, Scott (committee member).

Subjects/Keywords: Passive cameras; Video highlight detection; Snap point detection; Image segmentation; Video segmentation; Viewing panoramas

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Xiong, B. (2019). Learning to compose photos and videos from passive cameras. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://dx.doi.org/10.26153/tsw/5847

Chicago Manual of Style (16^th Edition):

Xiong, Bo. “Learning to compose photos and videos from passive cameras.” 2019. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://dx.doi.org/10.26153/tsw/5847.

MLA Handbook (7^th Edition):

Xiong, Bo. “Learning to compose photos and videos from passive cameras.” 2019. Web. 05 Jul 2020.

Vancouver:

Xiong B. Learning to compose photos and videos from passive cameras. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2019. [cited 2020 Jul 05]. Available from: http://dx.doi.org/10.26153/tsw/5847.

Council of Science Editors:

Xiong B. Learning to compose photos and videos from passive cameras. [Doctoral Dissertation]. University of Texas – Austin; 2019. Available from: http://dx.doi.org/10.26153/tsw/5847

10. Rawal, Aditya, Ph. D. in computer science. Discovering gated recurrent neural network architectures.

Degree: PhD, Computer science, 2019, University of Texas – Austin

URL: http://hdl.handle.net/2152/72839

► Reinforcement Learning agent networks with memory are a key component in solving POMDP tasks. Gated recurrent networks such as those composed of Long Short-Term Memory… (more)

Subjects/Keywords: Recurrent neural networks; Neuroevolution; Network architecture search; Meta-learning; Reinforcement learning; Language modeling; Music modeling

10 more images…

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Rawal, Aditya, P. D. i. c. s. (2019). Discovering gated recurrent neural network architectures. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/72839

Chicago Manual of Style (16^th Edition):

Rawal, Aditya, Ph D in computer science. “Discovering gated recurrent neural network architectures.” 2019. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/72839.

MLA Handbook (7^th Edition):

Rawal, Aditya, Ph D in computer science. “Discovering gated recurrent neural network architectures.” 2019. Web. 05 Jul 2020.

Vancouver:

Rawal, Aditya PDics. Discovering gated recurrent neural network architectures. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2019. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/72839.

Council of Science Editors:

Rawal, Aditya PDics. Discovering gated recurrent neural network architectures. [Doctoral Dissertation]. University of Texas – Austin; 2019. Available from: http://hdl.handle.net/2152/72839

11. -9199-0633. Continually improving grounded natural language understanding through human-robot dialog.

Degree: PhD, Computer Science, 2018, University of Texas – Austin

URL: http://hdl.handle.net/2152/68120

► As robots become ubiquitous in homes and workplaces such as hospitals and factories, they must be able to communicate with humans. Several kinds of knowledge… (more)

▼ As robots become ubiquitous in homes and workplaces such as hospitals and factories, they must be able to communicate with humans. Several kinds of knowledge are required to understand and respond to a human's natural language commands and questions. If a person requests an assistant robot to take me to Alice's office, the robot must know that Alice is a person who owns some unique office, and that take me means it should navigate there. Similarly, if a person requests bring me the heavy, green mug, the robot must have accurate mental models of the physical concepts heavy, green, and mug. To avoid forcing humans to use key phrases or words robots already know, this thesis focuses on helping robots understanding new language constructs through interactions with humans and with the world around them. To understand a command in natural language, a robot must first convert that command to an internal representation that it can reason with. Semantic parsing is a method for performing this conversion, and the target representation is often semantic forms represented as predicate logic with lambda calculus. Traditional semantic parsing relies on hand-crafted resources from a human expert: an ontology of concepts, a lexicon connecting language to those concepts, and training examples of language with abstract meanings. One thrust of this thesis is to perform semantic parsing with sparse initial data. We use the conversations between a robot and human users to induce pairs of natural language utterances with the target semantic forms a robot discovers through its questions, reducing the annotation effort of creating training examples for parsing. We use this data to build more dialog-capable robots in new domains with much less expert human effort (Thomason et al., 2015; Padmakumar et al., 2017). Meanings of many language concepts are bound to the physical world. Understanding object properties and categories, such as heavy, green, and mug requires interacting with and perceiving the physical world. Embodied robots can use manipulation capabilities, such as pushing, picking up, and dropping objects to gather sensory data about them. This data can be used to understand non-visual concepts like heavy and empty (e.g. get the empty carton of milk from the fridge), and assist with concepts that have both visual and non-visual expression (e.g. tall things look big and also exert force sooner than short things when pressed down on). A second thrust of this thesis focuses on strategies for learning these concepts using multi-modal sensory information. We use human-in-the-loop learning to get labels between concept words and actual objects in the environment (Thomason et al., 2016, 2017). We also explore ways to tease out polysemy and synonymy in concept words (Thomason and Mooney, 2017) such as light, which can refer to a weight or a color, the latter sense being synonymous with pale. Additionally, pushing, picking up, and dropping objects to gather sensory information is prohibitively time-consuming, so we investigate strategies for… Advisors/Committee Members: Mooney, Raymond J. (Raymond Joseph) (advisor), Stone, Peter (committee member), Niekum, Scott (committee member), Tellex, Stefanie (committee member).

Subjects/Keywords: Natural language processing; Human-robot dialog

10 more images…

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

-9199-0633. (2018). Continually improving grounded natural language understanding through human-robot dialog. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/68120

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Chicago Manual of Style (16^th Edition):

-9199-0633. “Continually improving grounded natural language understanding through human-robot dialog.” 2018. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/68120.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

MLA Handbook (7^th Edition):

-9199-0633. “Continually improving grounded natural language understanding through human-robot dialog.” 2018. Web. 05 Jul 2020.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Vancouver:

-9199-0633. Continually improving grounded natural language understanding through human-robot dialog. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2018. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/68120.

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

Council of Science Editors:

-9199-0633. Continually improving grounded natural language understanding through human-robot dialog. [Doctoral Dissertation]. University of Texas – Austin; 2018. Available from: http://hdl.handle.net/2152/68120

Note: this citation may be lacking information needed for this citation format:
Author name may be incomplete

12. Khandelwal, Piyush. On-demand coordination of multiple service robots.

Degree: PhD, Computer Science, 2017, University of Texas – Austin

URL: http://hdl.handle.net/2152/61382

► Research in recent years has made it increasingly plausible to deploy a large number of service robots in home and office environments. Given that multiple… (more)

▼ Research in recent years has made it increasingly plausible to deploy a large number of service robots in home and office environments. Given that multiple mobile robots may be available in the environment performing routine duties such as cleaning, building maintenance, or patrolling, and that each robot may have a set of basic interfaces and manipulation tools to interact with one another as well as humans in the environment, is it possible to coordinate multiple robots for a previously unplanned on-demand task? The research presented in this dissertation aims to begin answering this question. This dissertation makes three main contributions. The first contribution of this work is a formal framework for coordinating multiple robots to perform an on-demand task while balancing two objectives: (i) complete this on-demand task as quickly as possible, and (ii) minimize the total amount of time each robot is diverted from its routine duties. We formalize this stochastic sequential decision making problem, termed on-demand multi-robot coordination, as a Markov decision Process (MDP). Furthermore, we study this problem in the context of a specific on-demand task called multi-robot human guidance, where multiple robots need to coordinate and efficiently guide a visitor to his destination. Second, we develop and analyze stochastic planning algorithms, in order to efficiently solve the on-demand multi-robot coordination problem in real-time. Monte Carlo Tree Search (MCTS) planning algorithms have demonstrated excellent results solving MDPs with large state-spaces and high action branching. We propose variants to the MCTS algorithm that use biased backpropagation techniques for value estimation, which can help MCTS converge to reasonable yet suboptimal policies quickly when compared to standard unbiased Monte Carlo backpropagation. In addition to using these planning algorithms for efficiently solving the on-demand multi-robot coordination problem in real-time, we also analyze their performance using benchmark domains from the International Planning Competition (IPC). The third and final contribution of this work is the development of a multi-robot system built on top of the Segway RMP platform at the Learning Agents Research Group, UT Austin, and the implementation and evaluation of the on-demand multi-robot coordination problem and two different planning algorithm on this platform. We also perform two studies using simulated environments, where real humans control a simulated avatar, to test the implementation of the MDP formalization and planning algorithms presented in this dissertation. Advisors/Committee Members: Stone, Peter, 1971- (advisor), Grauman, Kristen (committee member), Niekum, Scott (committee member), Thomaz, Andrea (committee member), Veloso, Manuela (committee member).

Subjects/Keywords: Multi-robot coordination; Monte Carlo tree search; Markov decision processes; Probabilistic planning; Multi-robot systems

12 more images…

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Khandelwal, P. (2017). On-demand coordination of multiple service robots. (Doctoral Dissertation). University of Texas – Austin. Retrieved from http://hdl.handle.net/2152/61382

Chicago Manual of Style (16^th Edition):

Khandelwal, Piyush. “On-demand coordination of multiple service robots.” 2017. Doctoral Dissertation, University of Texas – Austin. Accessed July 05, 2020. http://hdl.handle.net/2152/61382.

MLA Handbook (7^th Edition):

Khandelwal, Piyush. “On-demand coordination of multiple service robots.” 2017. Web. 05 Jul 2020.

Vancouver:

Khandelwal P. On-demand coordination of multiple service robots. [Internet] [Doctoral dissertation]. University of Texas – Austin; 2017. [cited 2020 Jul 05]. Available from: http://hdl.handle.net/2152/61382.

Council of Science Editors:

Khandelwal P. On-demand coordination of multiple service robots. [Doctoral Dissertation]. University of Texas – Austin; 2017. Available from: http://hdl.handle.net/2152/61382

Open Access Theses and Dissertations