This Report is yet to be approved by a Coordinator.
Sonupriya's Level 1 AIML Report
24 / 10 / 2024
TASK 1: LINEAR AND LOGISTIC REGRESSION USING SCIKIT-LEARN
LINEAR REGRESSION:
Linear regression is a statistical method used to model the linear relationship between dependent variable(target) and one or more independent variable(features).It is commonly used for continuous outcomes like predicting house prices or sales. The goal is ti fit the straight line( y = mx + b ) to the data that minimizes the difference between the predicted values and actual values.
Here I have used diabetes dataset from scikit learn.
TASK 2: MATPLOITLIB
By using Matplotlib I've learnt creating interactive visualizations for the given data. It allows us to generate wide range of plots like, line charts, histograms, scatter plots and much more
TASK 3: NUMPY
Numpy is a library that provides wide range of mathematical functions to operate on large and multi-dimensional arrays and matrices.
Here is my appraoch to the given task.
TASK 4: METRICES
Metrics are quantitative measures used to assess the performance or effectiveness of a system, model, or process. In data analysis and machine learning, metrics help evaluate the accuracy, efficiency, and quality of algorithms or predictions.
There are two types of metrices:
1. REGRESSION METRICES
- Mean absolute error
- Mean sqaured error
- Root mean sqaured error
- R^2
Here i have used Placement CSV in LInear regression and calculated the above metrices
2. CLASSIFICATION/PERFORMANCE METRICES
- Confusion matrix
- Accuracy
- Precision
- Recall
Here i have used Heart CSV which is about chances of having a heart disease
TASK 4: LINEAR AND LOGISTIC REGRESSION FROM SCRATCH
LINEAR REGRESSION:
This task was helpful in understanding the backend of linear regression and mathematics behind it.(advertising dataset)
LOGISTIC REGRESSSION:
This task was helpful in understanding the backend of logistic regression and mathematics behind it.(breast cancer dataset)
TASK 6: KNN - K-NEAREST NEIGHBORS
K-Nearest Neighbors (KNN) is a simple and intuitive machine learning algorithm used for both classification and regression tasks. It works by identifying the 'k' nearest data points (neighbors) to a given query point and making predictions based on the majority class (for classification) or the average (for regression) of these neighbors.
Here i have built KNN model from scratch using the iris dataset.
Task 7:An elementary step towards understanding Neural Networks
This task has helped me understand the backend of aiml like how to train a model efficinetly and the complexity behind it.Though it was pretty much theoritical, I gained noteworthy information.
Understanding the basics of Deep Learning: CNN, ANN and RNN.
Deep learning has revolutionized various fields, from computer vision to natural language processing. At the heart of this technology are different types of neural networks, each designed for specific tasks. In this blog, we'll explore the three most prominent types of neural networks: Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN).
Artificial Neural Networks (ANN)
- Think of these as digital brain cells connected together
- They learn by looking at examples and adjusting themselves to get better
- Good for basic tasks like sorting emails or predicting prices
Convolutional Neural Networks (CNN)
- Specially designed to understand images
- Works by scanning images piece by piece to find patterns
- Great for tasks like recognizing objects in photos
Recurrent Neural Networks (RNN)
- These networks have a memory of what they've seen before
- Perfect for understanding things that come in sequences
- Used for tasks like translating languages or predicting what word comes next
The cool thing about all these networks is that they learn from examples, just like humans do. Each type is best suited for different kinds of tasks, making them super useful in today's technology.
How to Build GPT-4: A Super Simple Guide
- Collect Massive Data: Gather tons of text from books, websites, and articles
- Design the Brain: Use a Transformer architecture that can understand patterns in language
- Train the Model: Feed it all the data using thousands of powerful computers (GPUs)
- Fine-tune: Train it further on specific tasks to make it better at following instructions
- Test & Optimize: Make sure it's accurate and efficient
- Deploy: Launch it for people to use, with safety measures in place
Task 8:Mathematics behind machine learning
Thanks to the current situation I'm in, We have Fourier Transforms in M3 and MATLAB in our DSP lab, therefore this went pretty well.
Task 9: Data Visualization for Exploratory Data Analysis
I'm amazed with how interactive and efficient PLotly is! It was a great time learning PLotly and I'm curious to explore this much more.
Here I have built a 3d graph using the iris dataset
Task 10: An introduction to Decision Trees
Decision trees are flowchart-like structures that make predictions by splitting data based on different features, following a path from the root node to leaf nodes. They work like a series of "if-then" questions, making them easy to understand and visualize, though they can sometimes overfit to training data.
Task 11: SVM
Support Vector Machine (SVM) is a supervised ML algorithm that finds the optimal hyperplane to separate different classes in data. It maximizes the margin between classes to ensure robust classification and reduce overfitting. SVMs can handle non-linear data using kernel functions that transform data into higher dimensions. They excel in high-dimensional spaces and are commonly used for text classification, image recognition, and bioinformatics.
I'm so thankful to Marvel UVCE for opening their doors to me as an open learner - it helped me turn my AI/ML dreams into reality! Sohan Aiyyappa's and Dhruv's mentorship made this journey truly special, and I've grown so much along the way.