PRASHANTH RAVI PRAKASH
ac8t76@r.postjobfree.com 267-***-**** https://github.com/prashanth-prakash EDUCATION
University of Pennsylvania, SEAS Aug 2018 - Present Master of Science in Electrical Engineering
Relevant Courses: Machine Learning, Data Mining, Digital Signal Processing, Computational neuroscience GPA:3.43/4.0
Vellore Institute of Technology Jun 2013 - May 2017 Bachelor of Technology in Electrical and Electronics Engineering CGPA:8.38/10.0
RELEVANT EXPERIENCE
Research Assistant at Kording Lab Feb 2018 - Present Domain Approximation for EEG classification (Master’s Thesis)
Apply different algorithms to extract features from signals
Observe effects of ensemble learning techniques (XGBoost, Random Forests) on the features extracted
Explore transfer learning approaches using CNNs
Modified neural networks that are biologically plausible
Implemented algorithms that addressed different neural network architectures
Assisted in designing a new algorithm using Tensorflow
Submission of paper under review.
Graduate Teaching Assistant for Data Mining (ESE545) Aug 2018 - Dec 2018
• Designed projects and assignments for a class of 80 students
• Interacted with students to discuss concepts during weekly office hours Decode Evoked Potentials for Brain machine interface applications Oct 2016 - May 2017
• Used moving window techniques on the time series EEG data to counter noise
• Evaluated two correlation-based algorithms for the classification of brain states
• Published results in the IET Signal Processing Journal PROJECTS
Book Recommendation
• Analyzed conventional algorithms for collaborative filtering (SVD) for recommending books
• Applied Clustering techniques such as Kmeans, Kmeans++, to find similar users
• Implemented a neural network model for recommendations Sentiment Analysis on twitter dataset
• Stochastic Gradient descent using PEGASOS and ADAGRAD
• Achieved 84 % accuracy for ADAGRAD while 80% accuracy for PEGASOS Identification of spam
• Executed SVM for the detection of Spam using different kernels
• Achieved 85% testing accuracy using RBF kernel
• Used k-fold cross validation for hyper parameter optimization Classification of Images using CNN
• Programmed Convolutional Neural Networks to recognize images (CIFAR, MNIST dataset).
• Experimented with custom architectures on CIFAR-10 dataset Analysis of Click through rate data for recommending ads
• Executed deterministic algorithms to solve the multi armed bandit problem
• Achieved close to Zero-Regret using Upper Confidence Bound 1 algorithm Approximate retrieval on MovieLens dataset
• Implemented Locally sensitive hashing
• Reduced highly dimensional data to a lower dimension
• Found nearest neighbors using different similarity measures SKILLS
Languages: Python, SQL, MATLAB, C++
Tools: Numpy, csv, Tensorflow, Keras, Pandas, sklearn, scipy Additional: Problem Solving, Communication