On Going Activities
Development of Optimization based CNN algorithm
for glaucoma detection. The proposed algorithm will be
able to learn the network with fewer sample size. The
research aslo focused on developing neuaral network
architecture that can solve various limitations of
existing CNN models.
Development of a new evolutionary algorithm
based on the Bryophyllum tree. This will be a new
Optimization techinique the simulates formation of
Bryophyllum forest. The algorithm is going to keep track
of both maximization and minimization value of the
objective function simultaniously.
Development of a one-pass CNN learning approch
that needs no hyper-parameter tuning. The main objective
of this research is to develop a novel learning algorithm
for CNN that learns the covolutional and fully connected
layer kernels in one epoch.
More...
Synopsis of the latest publication
Meta-heuristic optimization schemes can
facilitate feature learning even with small amount of
training data. This paper presents a new feature learning
mechanism called multi-objective Jaya convolutional
network (MJCN) that attempts to learn meaningful features
directly from the images. The proposed scheme, unlike the
convolutional neural networks, comprises a convolution
layer, a multiplication layer, an activation layer and an
optimizer known as multi-objective Jaya optimizer (MJO).
The convolution layer searches meaningful patterns in an
image through the local neighborhood connections and the
multiplication layer projects the convolutional response
to a more compact feature space. (Read
more...)