Varun Ojha's Projects
Time Complexity of Algorithms
In data mining, anomaly detection (also outlier detection) is the identification of items, events or observations which do not conform to an expected pattern or other items in a dataset.
This is a lossy image compression scheme, in which an image is divided into non-overlapping blocks of pixels. For each block, threshold and reconstruction values are determined. The threshold is usually the mean of the pixel values in the block. Then a bitmap of the block is derived by replacing all pixels whose values are greater than or equal (less than) to the threshold by a 1 (0). Then for each segment (group of 1s and 0s) in the bitmap, the reconstruction value is determined. This is the average of the values of the corresponding pixels in the original block.
Backpropagation Neural Tree (BNeuralT): BNeuralT is an effective and robust algorithm for generating low complexity and high accuracy models for classification, regression and pattern recognition learning problems.
Complex-Valued and Multivalued Neural Network Implementation for researchers interested to work on this area.
Batch mode training of neural network by conjugate gradient method
Curse of Dimensionality Illustration
I have created this project hierarchy for Cython easy project organization. This package builds (a nice /probably the best example of) a Cython project hierarchy (organization) example. It facilitates creating module -> sub-module -> sub-sub-module. That facilitates a deep folder tree hierarchy.
These folders hold a repository of KNIME workflow for general and specific problems.
Physiological data processing, signal processing, and data fusion scripts and codes. ESUM: This research is a part of our research project "ESUM-Analysing tradeoffs between the energy and social performance of urban morphologies," at the Chair of Information Architecture, ETH Zurich, Zurich, Switzerland and the Chair of Computer Science for Architecture, Bauhaus University Weimar, Germany.
NSGA-II and GA: Implementation of Non-dominated Sorting Genetic Algorithm (NSGA-II) and Binary Genetic Algorithm for Combinatorial Optimization
Genetic Algorithm (GA) for making Ensemble of Predictors
Software: The Hierarchical Fuzzy Inference Tree Software Toolbox is a function approximation and feature selection tool that uses genetic programming for constructing a tree-like structure to construct an adaptive multi-layer perceptron. This standalone software toolbox solves prediction problems. The developed algorithm performs multiobjective in which it adaptively creates a simple model with high generalization ability.
Hypersphere search for optimizing space truss structure
Inferring Significant Links and Layers of Deep Neural Networks\\to Adversarial Attacks Using Synaptic Filtering
In pattern recognition, the k-nearest neighbour's algorithm (k-NN) is a non-parametric method used for classification and regression.[1] In both cases, the input consists of the k closest training examples in the feature space. The output depends on whether k-NN is used for classification or regression: https://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm
In statistics, the Kolmogorov–Smirnov test (K–S test or KS test) is a nonparametric test of the equality of continuous, one-dimensional probability distributions that can be used to compare a sample with a reference probability distribution (one-sample K–S test), or to compare two samples (two-sample K–S test).
Software: Non-Convex function optimization and meta-heuristic algorithm testing tool (a limited number of algorithms and functions have been implemented). Java codes of algorithms have been also included)
We propose an algorithm and a new method to tackle the classification problems. We propose a multi-output neural tree (MONT) algorithm, which is an evolutionary learning algorithm trained by the non-dominated sorting genetic algorithm (NSGA)-III.
Global optimization by multilevel coordinate search (MCS). This is a dividing rectangle (i.e., Brach and Bound) like an optimization algorithm. It is similar to the DIRECT algorithm.
Codes for regression through least square estimation and gradient decent training.
Software: This is predictive modelling software. In this file software, two training modes (batch and sequential) for classical backpropagation is implemented. All other hybrid-trainings such as the Ant Colony Optimization, Particle Swarm Optimization, Genetic Algorithm, Bacteria Foraging Optimization are implemented for batch mode training of Neural Network Only.
Software: The Adaptive Approximation Software Toolbox (Neural Tree Algorithm) is a function approximation and feature selection tool that uses genetic programming for constructing tree like structure to construct an adaptive multi-layer perceptron. This standalone software toolbox solves prediction problems. The developed algorithm performs multiobjective in which it adaptively creates simple models with high generalization ability.
Multi-Onjective and Many Objective Optimization. Python Implementation of NSGA 2 and 3 algorithms.
A package of Java codes: I wrote over 500 different types of programs during my Core JAVA learning.
Principal Component Analysis Java Code
This project package connects Python to MATLAB and vice-versa
The Sensitivity Analysis of Evolutionary Algorithms code repository provides a comprehensive framework to study the influence of EAs hyper-parameters. This code repository builds on two sensitivity analysis measures: elementary effect (MORISS METHOD) and variance-based effect (SOBOL METHOD).
A very basic C++ code for developing a self-organizing map (SOM) for data clustering (No GUI Visualization)
One vs All Class and Binary Class Support Vector Machine