Learning Goal Hierarchies from Structured Observations and Expert Annotations

Könik, Tolga; Laird, John E.

doi:10.1007/978-3-540-30109-7_17

Cited by 15 publications

(20 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The framework is based on time points and used only the successor temporal relation. Könik and Laird (2006) proposed a learning by observation framework to learn an agent program that mimics a human expert's behaviour in domains such as games. The learned concepts are used to generate behaviour rather than classification.…”

Section: Related Workmentioning

confidence: 99%

“…The important aspect to note for the above review is that most of the work in this area has been done on either artificial or simulated data (Moyle & Muggleton, 1997;Könik & Laird, 2006) or very simple real world data (Fern et al, 2002;Needham et al, 2005) that involves few objects, the events are of short duration and all the objects in the scene are involved in the events. In our case, the tracked data from videos is very large and at the same time more complex and noisy and contains more objects.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Learning Relational Event Models from Video

Dubba

Cohn

Hogg

et al. 2015

jair

View full text Add to dashboard Cite

Event models obtained automatically from video can be used in applications ranging from abnormal event detection to content based video retrieval. When multiple agents are involved in the events, characterizing events naturally suggests encoding interactions as relations. Learning event models from this kind of relational spatio-temporal data using relational learning techniques such as Inductive Logic Programming (ILP) hold promise, but have not been successfully applied to very large datasets which result from video data. In this paper, we present a novel framework remind (Relational Event Model INDuction) for supervised relational learning of event models from large video datasets using ILP. Efficiency is achieved through the learning from interpretations setting and using a typing system that exploits the type hierarchy of objects in a domain. The use of types also helps prevent over generalization. Furthermore, we also present a type-refining operator and prove that it is optimal. The learned models can be used for recognizing events from previously unseen videos. We also present an extension to the framework by integrating an abduction step that improves the learning performance when there is noise in the input data. The experimental results on several hours of video data from two challenging real world domains (an airport domain and a physical action verbs domain) suggest that the techniques are suitable to real world scenarios.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Learning Relational Event Models from Video

Dubba

Cohn

Hogg

et al. 2015

jair

View full text Add to dashboard Cite

show abstract

“…Human demonstrations have also received some attention to speed up reinforcement learning (Schaal 1996), and as a way of automatically acquiring planning knowledge (Hogg, Muñoz-Avila, and Kuter 2008), among others. Könik and Laird present a Relational Learning from Observation technique (Könik and Laird 2006) able to learn how to decompose a goal into subgoals, based on observing annotated expert traces. Könik and Laird's technique uses relational machine learning techniques to learn how to decompose goals, and the output is a collection of rules, thus showing an approach to learning planning knowledge from demonstrations.…”

Section: Related Workmentioning

confidence: 99%

An Ensemble Architecture for Learning Complex Problem-Solving Techniques from Demonstration

Zhang

Shrestha

Yoon

et al. 2012

ACM Trans. Intell. Syst. Technol.

View full text Add to dashboard Cite

We present a novel ensemble architecture for learning problem-solving techniques from a very small number of expert solutions and demonstrate its effectiveness in a complex real-world domain. The key feature of our "Generalized Integrated Learning Architecture" (GILA) is a set of integrated learning and reasoning (ILR) components, coordinated by a central meta-reasoning executive (MRE). The ILRs are weakly coupled in the sense that all coordination happens through the MRE. Each ILR learns independently from a small number of expert demonstrations of a complex task. During the performance, each ILR proposes partial solutions to subproblems posed by the MRE, which are then selected from and pieced together by the MRE to produce a complete solution. We describe the application of this novel learning and problem solving architecture to the domain of airspace management, where multiple requests for the use of airspaces need to be deconflicted, reconciled and managed automatically. Formal evaluations show that our system performs as well as or better than humans after learning from the same training data. Furthermore, GILA outperforms any individual ILR run in isolation, thus demonstrating the power of the ensemble architecture for learning and problem solving.

show abstract

“…A number of systems (e.g, van Lent & Laird, 1999;Wang, 1995;Konik & Laird, 2006) have also been developed to learn procedural rules or plan operators from observations of expert behavior. Wang's OBSERVER (Wang, 1995) learns STRIPS style operators; van Lent's KnoMic (van Lent & Laird, 1999) learns production rules for the Soar agent architecture and Konik's system (Konik & Laird, 2006) creates first order logic rules that are later converted into Soar productions.…”

Section: Learning By Observationmentioning

confidence: 99%

“…Wang's OBSERVER (Wang, 1995) learns STRIPS style operators; van Lent's KnoMic (van Lent & Laird, 1999) learns production rules for the Soar agent architecture and Konik's system (Konik & Laird, 2006) creates first order logic rules that are later converted into Soar productions. All three systems use similar behavior traces as our approach, although Wang's OBSERVER works only with primitive actions so there is no notion of non-atomic goals and thus no need to annotate them in the behavior traces.…”

Section: Learning By Observationmentioning

confidence: 99%

Behavior Bounding: An Efficient Method for High-Level Behavior Comparison

Wallace

2009

jair

View full text Add to dashboard Cite

In this paper, we explore methods for comparing agent behavior with human behavior to assist with validation. Our exploration begins by considering a simple method of behavior comparison. Motivated by shortcomings in this initial approach, we introduce behavior bounding, an automated model-based approach for comparing behavior that is inspired, in part, by Mitchell's Version Spaces. We show that behavior bounding can be used to compactly represent both human and agent behavior. We argue that relatively low amounts of human effort are required to build, maintain, and use the data structures that underlie behavior bounding, and we provide a theoretical basis for these arguments using notions of PAC Learnability. Next, we show empirical results indicating that this approach is effective at identifying differences in certain types of behaviors and that it performs well when compared against our initial benchmark methods. Finally, we demonstrate that behavior bounding can produce information that allows developers to identify and fix problems in an agent's behavior much more efficiently than standard debugging techniques.

show abstract

Learning Goal Hierarchies from Structured Observations and Expert Annotations

Cited by 15 publications

References 17 publications

Learning Relational Event Models from Video

Learning Relational Event Models from Video

An Ensemble Architecture for Learning Complex Problem-Solving Techniques from Demonstration

Behavior Bounding: An Efficient Method for High-Level Behavior Comparison

Contact Info

Product

Resources

About