Hello,
I am writing my SOP for Graduate Admissions in ECE. Many universities have a limit of 500 words or 4000 characters for such essays. I want to include a lot of specifics , about the work I have done , and the courses I have taken, but I cannot seem to include all of them . Here is my first draft of the SOP . I will include one paragraph at the end about the Professors I would like to work with at each University. I would like some help in trying to make the paragraphs appear more coherent , and also if possible to remove any unnecessary parts , so that I can include some other points.
%__________________________________________________________________
SOP
My broad research interests are in applying tools from areas such as signal processing, information theory and optimization , for modelling and extracting information from systems that arise in engineering. More specifically, I am interested in the design and analysis of algorithms for problems such as learning, dimensionality reduction and signal reconstruction from large scale data. In many cases, such as wireless sensor networks or networks of neurons in the brain, systems generating the data can be modelled as a complex network , built up of simple units having local interactions. Drawing upon ideas from fields like Information Theory ,Random Matrix Theory(RMT) and Graph Theory, I will like to work towards understanding the fundamental limits on the information processing capacity of such systems.
My first serious research experience was as a summer intern at Dr. A's lab at Univ.P in 2012, where I worked on detailed multicompartmental modelling of pyramidal neurons. Although my project work was mainly computational , I also learnt about the application of nonlinear dynamics in analysing neuronal spiking. After reading about the basics of bifurcation analysis from Strogatz's book , I was amazed at how the same mathematical tools can be used to analyse problems as diverse as the flashing behaviour of fireflies and firing of neurons in the brain. That summer , I also attended the Conf.# at Univ.Q. I particuarly liked the talk by Jeff Kirchmar on neuro-robotics. Influenced by my summer experiences, I decided to work on a survey of decoding, trajectory planning and control algorithms used in neural prosthetics for my final year UG project. While working on that project , I came up with a simple binning based spike train distance metric, which was later accepeted at an IEEE conference for publication.
After graduating from Univ.D , I continued to work in collaboration with Dr.A on neural spike train analysis. I proposed a new family of distance metrics , and performed a thorough comparison of the new measures with some existing similarity/distance measures in their ability to identify information rich features in spike trains.( Link1 ) I also worked on developing an area based operator, which can be used for extra cellular spike detection as well as feature extraction ,thus reducing the amount of hardware needed. (Under review at IEEE/ACM TCBB).
Having worked on some research projects, I realized that to make more significant research contributions , I needed to equip myself with more fundamental tools from Mathematics and Engineering. With that aim in mind, I joined the #### program at Univ.Q . .By taking courses on Linear Algebra, Probability Theory , Optimization ,as well as by teaching myself introductory analysis and topology from standard textbooks, I have been able to develop a firm understanding of some core areas in mathematics. To furthur expand my toolkit, I plan to take courses on Randomized Algorithms and Computational Complexity Theory next semester.
For my Masters thesis, I am working on developing an algorithm for reconstructing vectors from few measurements by nonconvex optimization. This algorithm uses a computationally efficient approximation of the Hessian of the objective function to find the search direction. Besides that, I have also been able to slightly generalize some of the existing results about the convergence of the algorithm . For my Computer Vision Course project , I have reformulated the Graph based TV minimization as a separable problem, and then applied D-ADMM for solving it.
I am writing my SOP for Graduate Admissions in ECE. Many universities have a limit of 500 words or 4000 characters for such essays. I want to include a lot of specifics , about the work I have done , and the courses I have taken, but I cannot seem to include all of them . Here is my first draft of the SOP . I will include one paragraph at the end about the Professors I would like to work with at each University. I would like some help in trying to make the paragraphs appear more coherent , and also if possible to remove any unnecessary parts , so that I can include some other points.
%__________________________________________________________________
SOP
My broad research interests are in applying tools from areas such as signal processing, information theory and optimization , for modelling and extracting information from systems that arise in engineering. More specifically, I am interested in the design and analysis of algorithms for problems such as learning, dimensionality reduction and signal reconstruction from large scale data. In many cases, such as wireless sensor networks or networks of neurons in the brain, systems generating the data can be modelled as a complex network , built up of simple units having local interactions. Drawing upon ideas from fields like Information Theory ,Random Matrix Theory(RMT) and Graph Theory, I will like to work towards understanding the fundamental limits on the information processing capacity of such systems.
My first serious research experience was as a summer intern at Dr. A's lab at Univ.P in 2012, where I worked on detailed multicompartmental modelling of pyramidal neurons. Although my project work was mainly computational , I also learnt about the application of nonlinear dynamics in analysing neuronal spiking. After reading about the basics of bifurcation analysis from Strogatz's book , I was amazed at how the same mathematical tools can be used to analyse problems as diverse as the flashing behaviour of fireflies and firing of neurons in the brain. That summer , I also attended the Conf.# at Univ.Q. I particuarly liked the talk by Jeff Kirchmar on neuro-robotics. Influenced by my summer experiences, I decided to work on a survey of decoding, trajectory planning and control algorithms used in neural prosthetics for my final year UG project. While working on that project , I came up with a simple binning based spike train distance metric, which was later accepeted at an IEEE conference for publication.
After graduating from Univ.D , I continued to work in collaboration with Dr.A on neural spike train analysis. I proposed a new family of distance metrics , and performed a thorough comparison of the new measures with some existing similarity/distance measures in their ability to identify information rich features in spike trains.( Link1 ) I also worked on developing an area based operator, which can be used for extra cellular spike detection as well as feature extraction ,thus reducing the amount of hardware needed. (Under review at IEEE/ACM TCBB).
Having worked on some research projects, I realized that to make more significant research contributions , I needed to equip myself with more fundamental tools from Mathematics and Engineering. With that aim in mind, I joined the #### program at Univ.Q . .By taking courses on Linear Algebra, Probability Theory , Optimization ,as well as by teaching myself introductory analysis and topology from standard textbooks, I have been able to develop a firm understanding of some core areas in mathematics. To furthur expand my toolkit, I plan to take courses on Randomized Algorithms and Computational Complexity Theory next semester.
For my Masters thesis, I am working on developing an algorithm for reconstructing vectors from few measurements by nonconvex optimization. This algorithm uses a computationally efficient approximation of the Hessian of the objective function to find the search direction. Besides that, I have also been able to slightly generalize some of the existing results about the convergence of the algorithm . For my Computer Vision Course project , I have reformulated the Graph based TV minimization as a separable problem, and then applied D-ADMM for solving it.