Scribd
Upload
2 views3 pages

Problem Domain

The document discusses various research papers focused on advancements in deep learning techniques, including convolutional and recurrent neural networks, optimization methods, and boundary detection improvements. It highlights the application of these techniques in areas such as mobile traffic management, conversational agents, financial data modeling, and semantic segmentation. The research emphasizes the importance of efficient learning algorithms and the development of flexible models for diverse tasks.

Uploaded by

Mridull551998
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
2 views3 pages

Problem Domain

The document discusses various research papers focused on advancements in deep learning techniques, including convolutional and recurrent neural networks, optimization methods, and boundary detection improvements. It highlights the application of these techniques in areas such as mobile traffic management, conversational agents, financial data modeling, and semantic segmentation. The research emphasizes the importance of efficient learning algorithms and the development of flexible models for diverse tasks.

Uploaded by

Mridull551998
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

Clustering of Research Papers

Problem Domain Research Highlights


Papers Number
 A large fraction of 3, 6, 25, 28 Convolutional and de-convolutional LSTM
Internet traffic is now recurrent networks, pruning, trained
driven by requests quantization, Huffman coding,
from mobile devices
with relatively small
screens and often
stringent bandwidth
requirements

 A DCNN designed 17,24, 27 Pure pattern-matching approach, DCNN


for long-term designed for long-term predictions, Monte
predictions. Carlo Tree Search (MCTS), RNNs
 Adapting Deep LSTM,
Convolutional Neural
Network training to
the task of boundary
detection can result
in substantial
improvements over
the current state-of-
the-art in boundary
detection

 To build intelligent 9, 12, 14, 23, End-to-end models, sparse prediction,


conversational 26, 39 LSTM, Regularizing RNNs,
agents
 A flexible technique 21 Determinantal Point Process (DPP),
for learning networks
with diverse neurons.
 Efficient optimization 11,29, RL of robust adversarial reinforcement
of a function. We first 30,31,33,35 learning (RARL0, LIPO, Optimization,
identify sufficient fairness of Learning Algo., Meta Networks,
conditions for the Genetic-Gated Networks (G2Ns), neural
consistency of programmer-interpreter (NPI),
generic sequential function-preserving transformations
algorithms and
formulate the
expected min, max
rate for their
performance
 Learning meta-level
knowledge across
tasks and shifts its
inductive biases via
fast parameterization
for rapid
generalization/
rapidly transferring
the information
stored in one neural
net into another
neural net. The main
purpose is to
accelerate the
training of a
significantly larger
neural net

 Comprehending and 32 Random Walk (RW)


modeling the Artificial Neural Network (ANN)
dynamics of financial
data

 Learner uses an 34 Iterative algorithm


iterative algorithm
and a teacher can
feed examples
sequentially and
intelligently based on
the current
performance of the
learner and design
algorithms to hep in
faster convergence.

 Broaden the reach of 37 CNNs, grid-like structures (e.g., images),


CNNs by developing Structure-Aware Convolutional Neural
structure-aware Networks (SACNNs)
convolution to
eliminate the
invariance, yielding a
unified mechanism of
dealing with both
Euclidean and non-
Euclidean structured
data.

 Models for semantic 16, 22 DNNs, Backpropogation Algorithm,


segmentation are
based on
adaptations of
convolutional
networks.
Developing a new
convolutional
network module that
is specifically
designed for dense
prediction

 Representation 2, 38, 40, Variational Gaussian process (VGP),


learning with deep 43,46, 47, 48 Bayesian nonparametric variational family,
generative models Generative Adversarial Imitation Learning
(GAIL),

 Residual learning 7, 40, 41,44 Residual learning framework, ImageNet


framework to ease CNNs, local Spatio-Temporal, Scene
the training of Recognition
networks that are
substantially deeper
than those used
previously.

You might also like