|
|||||||||||||
|
|
|
|||||||||||
Neural
Computation & Signal Processing Below, is a description of several projects which
are open for new MSc or PhD students. If you are interested, please contact
me regarding these projects or other projects that you would like to discuss.
Cardio/Pulmonary Functionality Inference The collection of projects are aimed at advancing
current state of the art in understanding cardiac functionality and its relation
to the sounds emanated from the heart. These projects relay on physiological
understanding of the cardio pulmonary system. This physiological support is
provided by Prof. Noam Gavriely Following algorithms and data manipulations
which result from the physiological process, our goal is to provide the
necessary machine learning and signal processing algorithms to best reflect
the cardiac functionality changes that we are mostly interested in. Data for
this work comes from healthy subjects as well as patients. Inference of basic cardiac functionality It is well known that there are strong
interactions between the (somatic) sensory system of the body and its brain
activity. In particular, we are interested in the interaction of brain
activity under high blood pressure and other cardiac malfunction. We are also
interested in the connection between the sensory system and epilepsy. First step in this project relies on concurrent
recording of EEG data and ECG data in epileptic patients. Students interested in this project should gain
basic knowledge in signal processing, learning machines and physiology and be
prepared to spend some of their time in a hospital environment. All the brain imaging related work are in
conjunction with Dr. Analysis of Heart Sounds Via HMM Related work by Ray Watrous. See also presentation
by Daniel Gil at my Advanced Seminar, as well as
background and publications of Guy Amit also in the advanced
seminar page. See also, Alex Weibel TDNN. Outline: Follow speech recognition
approach of extracting automatic features, vector quantization, HMM,
Segmentation and Clustering. Brain Imaging Analysis of Brain Imaging Data in Conjunction with Cardiac Activity It is well known that there are strong
interactions between the (somatic) sensory system of the body and its brain
activity. In particular, we are interested in the interaction of brain
activity under high blood pressure and other cardiac malfunction. We are also
interested in the connection between the sensory system and epilepsy. First step in this project relies on concurrent
recording of EEG data and ECG data in epileptic patients. Students interested in this project should gain
basic knowledge in signal processing, learning machines and physiology and be
prepared to spend some of their time in a hospital environment. All the brain imaging related work are in
conjunction with Dr. Analysis of
fMRI/EEG data: Emotional Effects This work is
done in collaboration with Dr. There are
subprojects in signal processing, machine learning and computer vision in
this subtopic. Computational Genomics Multi-database mining and graph mining algorithms Current knowledge about genomics and proteomics
is expanding rapidly with many databases being created for special purposes.
This project will draw information from a collection of different databases,
to obtain maximal amount of knowledge of specific genes (or proteins) with
respect to their effect on specific processes. The computational questions is
to determine those genes which are optimal targets for diagnosis or therapy,
namely those genes which participate in
a large number of pathways and processes (simpler problem) and for
therapy, those genes which when affected, block a certain pathway completely,
with minimal effect on other pathways. This is an NP hard problem which requires
development of novel methods. The work
will be in conjunction with leading researches in molecular biologists and
biochemistry to address some of the most current biological research
questions. Requirements: This project is intended for MSc
Students in the Bioinformatics program which have also good knowledge in
several database mining languages such as Python, knowledge in graph
theoretic methods and the specific use of the Graph Boost Library. Gene Dynamic
Network Inference using Bayesian Methods Related work: Hartemic This project is intended
for a MSc student. It intends to continue work done by Omer Berkman on
inference from a collection of “weak” Bayesian networks. The inference is
obtained on a regularitory network from a (long) time series of Genes (or
other markers) activations. In particular, the causal regulation is sought,
namely those markers which initiate the regulation of other markers. A similar causal effects are sought in brain
imaging inference, as we are using high resolution imaging (with EEG and
MEG). See above. Seismic Data Analysis Related work:
see presentation by Ido Yariv and Talmor at my Advanced Seminar. A recent project
was started on the infrasound properties of the mole rat and the way it
acquires information about the underground environment from infrasound. Mathematical
Properties of the Cross-correlation Function This is a
topic requires some knowledge in large deviation bounds, such as the Barankin
bound, and the Ziv Lemple bound. Some overview of the topic appears in Judah’s thesis. The goal is to
analyze the properties of the cross correlation function from multiple sonar
returns with the purpose of devising an optimal fusion of the information
that can be extracted from each of the cross-correlations. A simple paper on
this topic is Robust statistics from multiple pings improves
noise tolerance in sonar. Ultrasound
Image Enhancement using Multiple Pings This work
relies on the work that we did with enhancement of Sonar images using multiple
pings and attempts to apply the same concept for medical ultrasound. It
relies on the multiple pings idea (Robust statistics from multiple pings improves
noise tolerance in sonar) and robust motion estimation of
the ultrasound sensor (Multiple ping sonar accuracy improvement using
robust motion estimation and ping fusion). The goal is to
achieve a far more accurate ultrasound with less energy for lower damage to
the fetus. Neuronal
Optimal Coding Related work: Neuronal Goals: Efficient Coding and Coincidence
Detection. High
Dimensional Data Representation via Sound Related work: J. Berger and R. Coifman. This project is done in collaboration with Miri
Segal (PhD in Math and Visual & Audio Artist) and Assaf Talmudi (PhD in
Acoustics, and Musician) from the center for Digital Art in The idea is to provide acoustic information as an
additional aid to visual information and thus extending the number of free
dimensions which can be ‘observed’ concurrently. This is important when a lot
of information has to be analyzed together, for example a radiologist that
has to decide about a malignant tumor, can get additional about a wider
spectrum of the target via sound. Computer Systems and Networks of
biomedical sensors TinyOS and
Wireless Body Sensors Network TinyOS is an open-source operating system
designed for wireless embedded sensor networks. It features a component-based
architecture which enables rapid development. This OS has become a
standard in the recent development of a Wireless Body Sensors Network
and
tmote sky platform. We shall develop algorithms for real
time analysis of ECG using Pluto that is based on tmote sky. This
platform can handle up to six different body sensors at a wireless range of
125m. We shall also
develop software and algorithms to embed acoustic sensors into this platform.
Blue Tooth
Communication Based on the new CSR – BlueVoxFlash device,
http://www.csr.com/bluevoxflash/development.htm it is desired to develop a
system that can receive sensory input of one to three channels and send to
the computer for storage and analysis. Issues related to automatic gain,
codec (compression) have to be addressed. |
|
||||||||||||
|