Joseph M.

• Implemented Baum Welch Algorithm assuming Hidden Markov Model to recognize alphabets.
• Recorded alphabets manually and used MFCC to extract features, these were used to train the model.
• Obtained an accuracy of 98.2 % with an easy dataset and 97 % with a hard dataset.

• Developed a code in MATLAB to recommend movies to users based on their previous views and other customer’s views.

• Implemented collaborative filtering using knn and agglomerative clustering based on degree of closeness

• Different similarity measures such cosine similarity and Pearson’s similarity was used in both knn and agglomerative clustering, for data acquired from movielens data set which has been widely used for research purposes.

Achieved a RMSE of 0.9965 and… Show more

• Developed a code in MATLAB to recommend movies to users based on their previous views and other customer’s views.

• Implemented collaborative filtering using knn and agglomerative clustering based on degree of closeness

• Different similarity measures such cosine similarity and Pearson’s similarity was used in both knn and agglomerative clustering, for data acquired from movielens data set which has been widely used for research purposes.

Achieved a RMSE of 0.9965 and compared results against published papers and the results were presented in class. Show less

• Developed a code to simulate the Receiver Operating Characteristics curve of different detectors and determined the best detectors under various noise conditions and verified the results mathematically.

• Simulated the probability of false detection and probability of detection for different detectors to determine its ROC curve.

• Studied the performance of each detector under simulated conditions and compared their performance interns of Mean Squared Error (0.035) against a… Show more

• Developed a code to simulate the Receiver Operating Characteristics curve of different detectors and determined the best detectors under various noise conditions and verified the results mathematically.

• Simulated the probability of false detection and probability of detection for different detectors to determine its ROC curve.

• Studied the performance of each detector under simulated conditions and compared their performance interns of Mean Squared Error (0.035) against a predetermined statistic. Show less

• Implemented an FIR Low Pass filter with cut off frequency of 0.75π per sample to study their effects on a video and determined that it smoothens out the non moving objects in the video.

• Developed a code to determine the average speed of a circular moving object in any given video.

• Developed code to determine the centers of every moving object and used this knowledge and fps of the video to determine the average speed of the circular moving objects in the given video.

• Applied a map/reduce algorithm with Python on Hadoop to extract information such as subway arrival time, platform number and location from the large 2GB MTA(NY subway) dataset in the back end.
• Implemented Google Maps API with JavaScript to present NY subway route with the least delay of all passing stations in the front end.

• Developed an encryption algorithm using Logistic maps and authentication algorithm using fragile watermarks.
• Performed security analysis and results were demonstrated to panel.