T. Jebara, A. Pentland, "Action Reaction Learning: Analysis and Synthesis of Human Behavior," IEEE Workshop on the Interpretation of Visual Motion, 1998. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
Discriminative, Generative and Imitative Learning - Jebara (2002) (Correct)
....with an example below. Example: Joint versus Conditional Bayesian Inference In the following, we present a specific example to demonstrate this difference and to argue in favor of the conditional estimate p(yjx) versus the conditioned joint estimate (more details are in the Appendix of [97]) We demonstrate this with a simple 2 component 2D Gaussian mixture model with identity covariance and equal mixing proportions as shown in Figure 2.4(a) The likelihood for a data point z = x; y) is: 6 4 2 6 4 2 10 10 5 CONDITIONED JOINT DENSITY ESTIMATE CONDITIONAL DENSITY ESTIMATE ....
....describe learns an autonomous agent that is able to interact and respond appropriately to external stimulus from the world and participants within it. Given the ability to perceive real behavior in humans interacting in the world, we can collect data to learn a predictive model. In earlier work [97] Action Reaction Learning described a system that learns the behavior of two agents while they are interacting. This can be seen as a more specific instance of imitation where instead of having a teacher interacting with the world, we only consider two teachers interacting with each other. In ARL, ....
[Article contains additional citation context not shown here]
T.S. Jebara. Action reaction learning: Analysis and synthesis of human behavior. Master's thesis, MIT Media Laboratory, 1998. Vision and Modeling TR# 507.
Constructing Qualitative Event Models Automatically from.. - Fernyhough, Cohn, Hogg (1999) (Correct)
....object being tracked . Finally they provide a transition state model which specifies the possible transitions between the various actions. All this has to be manually provided and it is the recognition of the difficulty of providing such models which motivates our research. Jebara and Pentland [2] recognise the need not to have to tediously describe behaviour models directly but to enable a machine to learn them. In many ways the motivation of their work is similar to ours, but the underlying descriptive apparatus is entirely different: their time series predictions give probabilistic ....
Jebara A and Pentland A. Action reaction learning: Analysis and synthesis of human behaviour. In IEEE Workshop on The Interpretation of Visual Motion. http://vismod.www.media.mit.edu/žbobick/ivm-site/, 1998.
Action Reaction Learning: Automatic Visual Analysis and.. - Jebara, Pentland (1999) (18 citations) Self-citation (Jebara) (Correct)
....This forms a more discriminative model that concentrates modeling resources for the task at hand. L = N Y i=1 p(x i # y i j Theta) 4) We recently developed a variant of the EM algorithm called Conditional Expectation Maximization (CEM) for specifically optimizing conditional likelihood [10]. It essentially fits a probabilitydensity function (pdf) that maximizes the conditional likelihood of the response given the covariates. CEM is an iterativetechnique which uses fixed point solutions (i.e. as opposed to gradient descent) to converge the parameters of a conditional densitytoa ....
.... c ) Figure 8: Conditional Density Estimation for CEM and EM properties are critical for a good conditional density p(yjx) In regression experiments on standardized databases, mixture models trained with CEM outperformed those trained with EM as well as conventional neural network architectures [10]. Thus, the CEM algorithm is used to estimate the conditional probability density (cpdf) relating past time series sequences (x) to their immediate future values (y) from training data (thousands of x# y pairs) A total of M Gaussians are fit to the data as a conditioned mixture model. This is ....
[Article contains additional citation context not shown here]
T. Jebara. Action-reaction learning: Analysis and synthesis of human behaviour. Master's thesis, Massachusetts Institute of Technology,May1998.
Ontology and Taxonomy Collaborated Framework for Meeting.. - Hakeem, Shah (2004) (Correct)
No context found.
T. Jebara, A. Pentland, "Action Reaction Learning: Analysis and Synthesis of Human Behavior," IEEE Workshop on the Interpretation of Visual Motion, 1998.
Discriminative, Generative and Imitative Learning - Jebara (2002) (Correct)
No context found.
T.S. Jebara. Action reaction learning: Analysis and synthesis of human behavior. Master's thesis, MIT Media Laboratory, 1998. Vision and Modeling TR# 507.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC