Adaptive multi-modal human behavior detection framework
Developed an adaptive multi-modal human behavior detection and classification framework. The system is designed to identify emotional state based on a variety of cues, including typing patterns, speech tone, facial expressions, and body movements. The primary goal of this project was to research ways of using Dempster–Shafer belief functions in multi-class support vectors employed in a fuzzy logic accounting for various behavioural performance matrices human subjects to learn and classify mood. Publication: Jul 2016 International Journal of Artificial Intelligence Paradigm (IJAIP)
Machine Learning Scientist for past 5years! Programming Languages: Python | C++ | JAVA SE | C# | Assembly Databases: MySQL | MongoDB | Spark SQL | Cassandra | BigQuery Machine Learning Frameworks: Scikit-learn | TensorFlow | Keras | Caffe/Caffe2 | Torch/PyTorch | Cuda | OpenCV | DLib | MATLAB BIGData Frameworks: Hadoop | SparkMLib/SparkML | Spark Streaming | Kafka | Kubernetes | Elasticsearch | Docker Version Control: Git, SVN Domain expertise in ML: Regression (Linear, Logistic, Regularized models - Ridge/Lasso), Decision Trees, Hypothesis Testing (T-test, Chi-Square, Wilcoxon, Cramers-V, Anova etc.), Clustering(K-Means, Hierarchical) , PCA/Factor Analysis, SVM, Random Forest and Decision Tree modules, Gradient Boosting, AdaBoost, XGBoost, Monte Carlo simulations, Deep Learning/Neural Networks (CNNs, RNNs), Stacking/Ensembles, Linear Blending, Grid Search, SGD classifier, Kalman Filter, EM Algorithm, Linear/Stochastic Programming, Denoising Autoencoders, Adagrad, Adam.