Roushan Kumar

Abstract This paper proposes a unified moments matching (MM) control technique for the regulation of output voltage of non-minimum phase (NMP) system, with applications to various orders of DC–DC boost converters. While prior studies use tuning techniques that are converter-specific and robust control approaches that are complicated, the proposed method uses a simple proportional-integral-derivative (PID) controller designed based on MM principles.

Roushan Kumar · Follow 3 min read · -- Imagine building a K-NN system that classifies handwritten digits, recommends products, and predicts house prices — all while handling millions of data points efficiently. In this final blog of our K-NN series, we’ll stitch together everything we’ve learned (distance metrics, KD-Trees, LSH, cross-validation) into a production-ready pipeline. We’ll use the MNIST dataset for classification, California housing data for regression, and optimize for scalability.

Roushan Kumar · Follow 3 min read · -- Imagine trying to find your look-alike in a crowd of 8 billion people. Checking every face would take a lifetime. Locality Sensitive Hashing (LSH) solves this by grouping similar items into “buckets,” reducing search times from days to milliseconds. In this blog of our K-NN series, we’ll explore how LSH works, implement it for cosine/Euclidean similarity, and apply it to massive datasets.