-
Notifications
You must be signed in to change notification settings - Fork 0
/
interests.html
93 lines (85 loc) · 7.64 KB
/
interests.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
<!DOCTYPE html>
<!--[if (gte IE 9)|!(IE)]><!--><html lang="en"> <!--<![endif]-->
<head>
<style>
a {color:blue;}
</style>
<meta charset="utf-8">
<title>Prashanth Balasubramanian</title>
<link rel="shortcut icon" href="iconsys.ico" />
<!-- Mobile Specific Metas
================================================== -->
<meta name="viewport" content="width=device-width, initial-scale=1, maximum-scale=1">
<!-- CSS
================================================== -->
<link rel="stylesheet" href="css/zerogrid.css">
<link rel="stylesheet" href="css/style.css">
<link rel="stylesheet" href="css/lightbox.css">
<!-- Custom Fonts -->
<link href="font-awesome/css/font-awesome.min.css" rel="stylesheet" type="text/css">
<link rel="stylesheet" href="css/menu.css">
<script src="js/jquery1111.min.js" type="text/javascript"></script>
<script src="js/script.js"></script>
</head>
<body>
<div class="wrap-body">
<div class="header">
<div id='cssmenu' >
<ul>
<li class="active"><a href='index.html'><span>Home</span></a></li>
<li><a href='interests.html'><span>Interests and Future Work</span></a>
</li>
</li>
<li><a href='cv.html'><span>CV</span></a></li>
<li><a href="https://in.linkedin.com/in/prashanth-balasubramanian-b043a0110"><span>Linkedin</span></a></li>
<li><a href="https://github.com/prashb94"><span>GitHub</span></a></li>
<li class='last'><a href='contact.html'><span>Contact</span></a></li>
<li class='crypto'><a href='crypto.html'><span>Cryptocurrency</span></a></li>
</ul>
</div>
</div>
<section id="container">
<div class="wrap-container">
<section class="content-box boxstyle-1 box-1">
<div class="zerogrid">
<div class="row wrap-box">
<div class="col-1-2">
<div class="wrap-col">
<center>
<img src="images/interest.png">
</center>
</div>
</div>
<div class="col-1-2">
<div class="wrap-col">
<div class="row">
<div class="t-center" style="padding-top: 60px;">
<div class="header">
<p><strong><u>Newest:</u></strong></p>
After an incredible summer at Tsinghua University and a pretty intense time in Germany, I've just begun my Master's in CS at ETH Zurich.
<p></p>
<p>I've just been accepted into the Big Data Summer Program at Tsinghua University that is set to take place from July 10 - July 30 and am really looking forward to it!</p>
<p><strong><u>My work in Germany!</u></strong></p>
<p>It's been almost 4 months since I began my work here in Germany as a research intern at the L3S Research Center. The level of work has been incredible and I've been rapidly learning new concepts that I did not expect to learn for a few more years atleast. Currently, I'm working on a tool called ArchiveSpark based on the Apache Spark framework, developed by my supervisor Helge Holzmann. So far, I've learnt about Latent Semantic Analysis (LSA) and successfully managed to analyze web archives using LSA. The next step is to clean the data, and am currently researching existing methods and different frameworks available, to achieve the most efficient possible duplicate detection(LSH, Shingling etc.).</p>
<p>By the end of my tenure here, my goal is to have applied ArchiveSpark and incorporated into it a novel framework that can achieve efficient information extraction from large data sets</p>
<p><strong><u>A basketball playing bot that never misses a shot-</u></strong></p>
<p>While playing basketball a few weeks ago, I was thinking about how interesting it would be to watch a bot shoot a basketball from any point on the court and never miss. After doing some research, I believe that if I am able to create such a bot, not only will it open up a whole new era of possibilities, but also disseminate novel methods of machine learning. As Dr. Fei-Fei Lin remarkably trained a computer to replicate the intellectual capacity of a three year old child, would it not be remarkable to replicate the intellectual capacity of a sports prodigy?</p>
<p><u>My theory:</u> I propose building a simple robotic shooting arm positioned at about 6.5ft above ground level, guided by microcontroller neural nets.For example, for shots within the half court range, publicly available are more than 1 million hours of videos of NBA players making all possible types of shots. But, beyond the half court range, the amount of training material is highly limited as these shots are impractical. Thus, for the second category of baskets, would a transductive learning algorithm coupled with basic physics trajectory motion formulae (Much like the ones I used to create "Falling Balls"- <a href='cv.html'><u>CV</u></a>), be as efficient as a regular induction type algorithm?</p>
<br>
<p>Having witnessed and been a part of the horrid traffic conditions in India, my largest goal is to perfect the autonomous car. I do not exaggerate when I say that the traffic conditions in India are analogous to the Brownian motion. Millions of lives are lost yearly due to reckless driving and poor driving conditions, not only in India, but in most developing countries. My papers on the APAA system and Crash Analysis system will be published sometime next year after the patent specifications are complete, and these delve deeper into these type of conditions and offer some sort of an initial feasible solution.</p>
<p>I not only wish to curb these numbers, but I am just absolutely fascinated by the hypothesis of a self-driving vehicle that is not tailored to a particular state or city, but something that can be deployed anywhere in the world. Well the mars rover definitely doesn't have to worry about traffic.<p>
<p>My research interests lie in deep learning, parallel programming, embedded systems and machine learning. Alex Krizhevsky's <a href='http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf'><u>paper</u></a> on classification and training on multiple GPU's gave me several things to think about and I've come up with a whole bunch of interesting theories about learning methods that I wish to test out and possibly implement during my Master's degree/Ph.D. Over the next 5 or so years I wish to dedicate all my time and efforts to research and development to fabricate the ultimate autonomous vehicle, among several others. I find it not bigoted to focus only on autonomous vehicles, as focusing on such a topic involves traversing a vast array of disciplines, each of which I'm sure will be conducive to my future as a figure in academia.</p>
<p>I'm just an undergraduate student at the moment, with an overly keen interest in Artificial Intelligence and I end up spending a lot of my time reading papers and technical reports that can possibly enrich my knowledge. I'm still quite far off from understanding the mysteries of Semi-Supervised Learning and Transductive Learning but I found several parts of this Ph.D. thesis quite enlightening - <a href='http://www.cs.technion.ac.il/~pechyony/phd_thesis.pdf'><u>Theory and Practice of Transductive Learning</u></a></p>
<p>I will continue posting my future interests and works on this page.</p>
</div>
</div>
</div>
</div>
</div>
<a href="http://m.maploco.com/details/cf35djat"><img style="border:10px;" src="http://www.maploco.com/vmap/s/9029621.png" alt="Locations of Site Visitors" title="Locations of Site Visitors"/></a>
</div>
</section>
<script src="js/lightbox-plus-jquery.min.js"></script>
</div>
</body>
</html>