Search this site

Embedded Files

Home
DL unit 1
Advanced AI
Data Mining
Data Mining Lab
Reinforcement Learning
Additional Reading
Syllabus

Home
DL unit 1
Advanced AI
Data Mining
Data Mining Lab
Reinforcement Learning
Additional Reading
Syllabus
More

DL unit 1

Artificial Neural Networks (ANNs) are computational models inspired by the structure and functioning of biological neural networks, such as the human brain. ANNs are a subset of machine learning algorithms that aim to recognize patterns and make decisions in a manner similar to the human brain. They consist of interconnected nodes (artificial neurons) organized into layers.

Models of a Neuron:

The basic building block of an artificial neural network is the artificial neuron, also known as a perceptron. A perceptron takes multiple input values, applies weights to them, sums them up, and then passes the result through an activation function to produce an output. The structure of a basic artificial neuron can be represented as follows:

Here's a brief overview of the components:

Inputs (x1, x2, ..., xn): These represent the input features or signals that the neuron receives.
Weights (w1, w2, ..., wn): Each input is associated with a weight, which determines the strength of the connection between the input and the neuron.
Summation Function (Σ): The weighted inputs are summed up: Σ(wi * xi).
Activation Function (f): The summation result is passed through an activation function, which introduces non-linearity to the model and determines the output of the neuron.
Output (y): The final output of the neuron after the activation function is applied.

Activation Functions:

Activation functions play a crucial role in artificial neural networks by introducing non-linearities. They determine whether a neuron should be activated or not based on the weighted sum of inputs. Here are some commonly used activation functions:

Step Function: It produces a binary output (0 or 1) based on a threshold. However, it is not commonly used in hidden layers of deep neural networks due to its lack of differentiability.
Sigmoid (Logistic) Function: It squashes the output between 0 and 1, making it suitable for binary classification problems. The formula is f(x) = 1 / (1 + e^(-x)).
Hyperbolic Tangent (tanh) Function: Similar to the sigmoid but maps the output between -1 and 1. It helps mitigate the vanishing gradient problem. The formula is f(x) = (e^(2x) - 1) / (e^(2x) + 1).
Rectified Linear Unit (ReLU): It replaces negative values with zero and is widely used in hidden layers due to its simplicity and effectiveness. The formula is f(x) = max(0, x).
Softmax Function: It is often used in the output layer for multi-class classification problems, as it converts a vector of real numbers into a probability distribution.

These activation functions contribute to the neural network's ability to model complex, non-linear relationships in data. The choice of activation function depends on the specific requirements of the task at hand.

Updated

Unit-3 Deep Learning and advanced algorithms track: Introduction, Performance measure, Transfer learning, Autoencoders, Natural Language Pr ocessing (NLP)

Back propagation with RNNs

Natural Language Processing & Word Embeddings

Introduction to Word Embeddings

Word Representation

Using word embeddings

Properties of word embeddings

Embedding matrix

Learning Word Embeddings: Word2vec & GloVe

Learning word embeddings

Word2Vec

Negative Sampling

GloVe word vectors

Applications using Word Embeddings

Sentiment Classification

Sequence models & Attention mechanism

Various sequence to sequence architectures

Basic Models

Picking the most likely sentence

Beam Search

Refinements to Beam Search

BLEU Score

Attention Model Intuition

Attention Model

Attention model, Unsupervised Learning, Representation Learning, Generative Adversarial Networks, Bayesian a pproaches to machine learning and deep learning, GANs, Rein forcement Learning .

Resources:

Sequence model attention

Attention Model

Attention and Transformer ( Machine Translation + Attention is part of the syllabus only)

https://github.com/ashishpatel26/Andrew-NG-Notes

Lab Experiments:

Introduction of Python and Linear algebra:

Task:

· understanding of variables, functions and constant in python

· Loop structures and if else structure

· write & execute Python programs using Interactive mode and script mode

· Usage of Array, List, tuples, and dictionary.

· Matrix operations (Addition, Multiplication, Transpose)

Understanding Data:

Task:

· Pandas Data structure Series and Dataframe

· Analysis of Dataframe

· Loading of CSV file

· Indexing and Labeling

· Basic operations with files

· Some Pre defined functions(head(),tail(), info(), eye() and Many More)

Dataset Used: titanic.csv dataset

Data Preprocessing and Correlation Matrix:

Task:

· Handling the missing values

· Manipulation of Rows/attributes

· Adding and Deleting columns

· Standardization and Normalization

· find correlation among attributes

· Visualization Techniques (Matplotlib Library)

Dataset Used: pima-indians-diabetes.data.csv

Data Visualization:

Task:

· Line chart

· Bar chart

· histogram

· pie chart

· Scatter chart

Dataset Used: Company_Sales_Data.csv

Logistic Regression, KNN, Linear Regression:

Lab:

· Perform EDA

· Apply the algorithm on dataset

Dataset Used: Diabetes.csv (classification), USA_Housing_price.csv

Decision Trees, Random Support Vector Machine & Naïve Bayes:

· Apply different algorithms on dataset.

· Compare all the result using plot

Dataset: breast_cancer_data.csv

Clustering (K means clustering and DBscan):

· Apply different algorithms on dataset.

· Compare all the result using plot

Dataset: driver_data.csv

Assignment:

1. Write code for the following operations:

§ Create numpy array size of 5*5

§ Calculate the sum of all the elements of array.

§ Calculate the sum of all the diagonal elements.

2. Write the python code for salary dataset.

§ What is the average BasePay?

§ What is the highest amount of OvertimePay in the dataset?

§ How much does JOSEPH DRISCOLL make (including benefits)?

§ What is the name of highest paid person (including benefits)?

§ What was the average (mean) BasePay of all employees per year? (2011-2014)?

3. Visualize the following data on company_sales_data.csv.

§ Read Total profit of all months and show it using a line plot

§ Get total profit of all months and show line plot with the following Style properties

§ Read all product sales data and show it using a multiline plot

4. The Dataset contains information about users on a Social Networking site and using that information as features for our ML model, the model predicts whether a particular user after clicking on an ad on the Social networking site goes on to buy a particular product or not. It is a CLASSIFICATION PROBLEM as the output says whether the user buys the product or not, so it’s either a yes or a no. Well this particular Social Network has a Business client which lets assume is a car company which advertises it by putting ads on the social networking site. Now the work of the social network here is to gather information as to whether the user bought the product or not. Apply the following algorithm:

§ KNN algorithm

§ Support Vector Machine

§ Decision tree

§ Random forest

§ Naïve base classifier

5. Apply K clustering and DBscan algorithm for credit card dataset. The sample dataset summarizes the usage behavior of about 9000 active credit card holders during the last 6 months. The file is at a customer level with 18 behavioral variables.

6. We have a data which classified if patients have heart disease or not according to features in it. We will try to use this data to create a model which tries predicting if a patient has this disease or not. Use the following algorithms:

§ KNN algorithm

§ Support Vector Machine

§ Decision tree

§ Random forest

§ Naïve base classifier

Download the datasets

Deep Learning Marks- 2022

Reinforcement Learning Assignment (last date 04-10-24)

Report abuse

Page details

Page updated

Report abuse