Page 27 - profile2014.indd
P. 27
Multimedia Technologies Lab
stage-wise closed-form solutions to accomplish the learn- 2. Extending analysis operator learning to supervised learning
ing task. In addition, our research output for speeding up the to yield more effective features. At the same time, we are ad-
sparse fast Fourier transform (sFFT) is promising. We propose dressing how to formulate a unified approach to both analysis
a novel sFFT that exploits downsampling in the time domain. and synthesis operator learning.
The resulting algorithm is more efficient and easy to implement,
while yielding comparable results to those obtained with the Studying how to sample Fourier measurements directly with-
original sFFT. in the framework of compressive sensing for fast recovery. As
the theoretical recovery bound and practical performance for
Through our ongoing research, we will continue to investigate sparse signal recovery algorithms are still not consistent, our
the aforementioned topics more thoroughly. In particular, we research aims to close the gap between them.
intend to focus on:
1. Studying the signal separation problem, by leveraging with
an analytical model that allows each component signal to be
sparsely analyzed by only one dictionary. We are also consider-
ing the more general problem of structure sparse representa-
tion, and expect to gain a deeper understanding of its proper
modeling.
Figure: We extract the joint positions and movements from a clip (shown in the upper-left and lower-left figures). The information is used
to construct a complete body skeleton (shown in right figure). It is thus feasible to manipulate the original pose by leveraging with inverse
kinematics techniques.
27