Dive into my treasure trove of notable projects here!
This project uses the Cambridge Labeled Objects in Video dataset, comprising 101 images (960x720 pixels).
Each pixel is manually classified into one of 32 object classes relevant to driving, with an extra "void" label for ambiguous areas.
The dataset includes a color-class association file (label_colors.txt) and the images are in 24-bit color PNG format.
For every original frame, there's a labeled counterpart with an "_L" added to the filename.g.
Embark on a thrilling journey through my project - 'Mastering Deep Reinforcement Learning using the OpenAI Gym Environment'.
Here, I've unleashed the power of artificial intelligence, implementing complex reinforcement learning algorithms in a dynamic and challenging setup.
Explore how I've navigated this digital labyrinth and unlocked the potential of AI-driven decision-making.
In this Project, I tried solving the classical reinforment learning problem, balancing the exploration and exploitation dilemma.
This project showcases my approach to the age-old explore-exploit dilemma. I've used a multi-armed bandit algorithm to strike a balance between exploiting
known rewards and exploring new possibilities, providing an innovative solution to classic reinforcement challenges.
This project focuses on binary classification using various machine learning algorithms.
The dataset is manipulated and visualized using libraries such as NumPy, Pandas, Matplotlib, and Seaborn.
The classification models used include Logistic Regression, Decision Tree Classifier, and Random Forest Classifier.
This venture saw me using Python to implement diverse clustering algorithms, showcasing their power and precision when applied to real-world data.
Tasked with clustering words, I navigated the challenges of linguistic data, unveiling patterns and drawing meaningful conclusions.
Discover the intriguing world of unsupervised learning and its applications in this project.
The Perceptron algorithm is a basic neural network architecture for binary classification. It contains two layers of nodes, input and output, connected by weighted connections.
The algorithm aims to find a linear decision boundary to separate two classes by adjusting the weights and bias term.
The activation function of the perceptron is a signed linear function, and learning is achieved by modifying the weights when the predicted label does not match the true label.
In this project, I implemented the binary perceptron algorithm from scratch
In this project I designed and implemented several voting rules. In a voting setting, we have a set of n agents and a set of m alternatives. Every agent has a preference ordering ≻ where α ≻ β means that the agent prefers alternative α to alternative β. A preference profile is a set of n preference orderings, one for every agent. For example, if we have a voting setting with 4 agents and 4 alternatives, one possible preference profile could be the following: Agent 1: α≻γ≻β≻δ Agent 2: α≻β≻δ≻γ Agent 3: γ≻β≻α≻δ Agent 4: β≻α≻δ≻γ A voting rule is a function that takes as input the preferences of a set of agents and outputs a winning alternative..
I implemented a simple bank accounting system. The code defines two classes, BasicAccount and PremiumAccount, representing bank accounts with basic and premium features, respectively.
They both have methods for depositing, withdrawing, and getting account information. PremiumAccount allows for overdrafts, with an overdraft limit set by the account holder.
An instance of PremiumAccount is created with an initial balance of 1000.00 and an overdraft limit of 20000.00, and the current balance is printed.
In this project I designed and implemented a simple catche management app. The code implements two cache management policies, FIFO and LFU, to manage a cache of fixed size. It takes input page requests from the user and checks if each page is already in the cache. If it's a miss, the code selects a page to evict based on the policy, either the first-in page for FIFO or the least frequently used page for LFU. The cache is printed at the end of each policy's run.
Peek into my exploration with the project titled 'Unraveling the EMNIST Balanced: A Dive into Handwritten Character Recognition'.
This project utilizes the EMNIST Balanced dataset - an extension of the famed MNIST, housing over 131,000 images of handwritten characters.
I've employed this balanced dataset to ensure unbiased model training for tasks like character recognition and OCR.
With this project, witness how I've used PyTorch to navigate the fascinating world of handwriting recognition.
In this analysis I employed advanced natural language processing (NLP) techniques to thoroughly explore and understand the content of a large collection of tweets written in Dutch.
In this Project, I analyzed the trends of Bitcoin and Tether (USDT) in order to gain insights into the current state of the cryptocurrency market. The analysis was conducted using two datasets: daily price and volume data for Bitcoin and transaction data for Tether. These datasets were combined to produce a dataset containing daily Bitcoin prices and daily USDT transaction volumes for further analysis.
In this project, I explored the findings from an analysis of HubPay customer data, focusing on descriptive analytics, data preparation and cleansing, and the development of classification models to predict churn.
#