diff options
| -rw-r--r-- | finale/sections/intro.tex | 12 |
1 files changed, 6 insertions, 6 deletions
diff --git a/finale/sections/intro.tex b/finale/sections/intro.tex index 0806c98..4d59ac6 100644 --- a/finale/sections/intro.tex +++ b/finale/sections/intro.tex @@ -22,18 +22,18 @@ Specifically: \begin{itemize} \item we propose a Bayesian Inference formulation of the NIP problem in the Generalized Linear Cascade (GLC) Model of \cite{pouget} and show how to apply - MCMC and variationel inference to it. - \item we show how to leverage this Bayesian formulation to design active - learning heuristics where the experimenter is able to dynamically + MCMC and Variational Inference to it. + \item we show how to leverage this Bayesian formulation to design Active + Learning heuristics where the experimenter is able to dynamically choose the source node at which the observe cascades originate. - \item we show empirically that active learning greatly improves the speed - of learning compared to i.i.d observations. + \item we give empirical evidence that Active Learning greatly improves the + speed of learning compared to i.i.d observations. \end{itemize} The organization of the paper is as follows: we conclude this introduction by a review of the related works. Section 2 introduces the notations and the Generalized Linear Model, Section 3 presents our Bayesian Inference -formulation. The active learning approach is described in Section 4. Section +formulation. The Active Learning approach is described in Section 4. Section 5 gives our experimental results. Finally we conclude by a discussion in Section 6. \input{sections/related.tex} |