Sie sind auf Seite 1von 15

MACHINE LEARNERS

Artificial Intelligence &


Machine Learning
Bootcamp

Course Content
About this bootcamp:
This is time for the next technological evolution. The era when the fifth generation of computers is
finally getting closer to its end. Now, it is the era of Artificial Intelligence, and it is necessary to adopt
the skill as soon as possible to stay ahead from others in the field.In this bootcamp you’ll learn valuable
artificial intelligence skills that are in demand across countless industries. Machine Learning is a
first-class ticket to the most exciting careers in AI. This bootcamp will take you from simple machine
learning to the futuristic Deep Learning along with Reinforcement learning. Learn to work on computer
vision and natural language processing along with creating multi-models to complete tasks like image
captioning and video summarization.During the bootcamp you will be implementing all these break
through technologies using high end libraries like Scikit-learn, Keras, Tensorflow and many more.

Learn the Cutting-Edge Techniques:


In this course you will get to learnstate of the art algorithms to create models that replicate human
brain: Algorithms like Deep Learning and Q-Learning will be the core of this bootcamp.

Cloud Computing:
Take advantage of Google Colab's GPUs and Kaggle’s free GPU to train your neural network faster.

Five Star Rating:


You will be assigned five classical machine learning (ML) and deep learning (DL) problems as assign-
ments. Submission of each project successfully will increase your rating by a star. Be prepared for this
awesomeness.

Machine Learning for Recommender systems


Solve Titanic Survival using NN
Object Detection using Deep Learning
Text Summarization Using Sentence Embeddings and/or Image Captioning in Deep Learning
Build a Game Bot

Core Skills:

Machine Learning
Supervised Learning
Unsupervised Learning
Computer Vision
Deep Learning
Natural language processing
Reinforcement Learning

Prerequisite Knowledge :
Algebra, Calculus & Python — Make sure to set adequate time on your calendar for focused work. This
program has been designed specifically for students who are interested in Artificial Intelligence, Machine
Learning, and/or Deep Learning, and who have knowledge of Python programming and basic statistics. In
order to succeed in this program, we recommend having intermediate experience with Python, including
numpy and pandas, and basic knowledge of machine learning. You’ll also need to be familiar with algebra
and calculus.

MACHINE LEARNERS
Learn Complete AI Toolkit:
Learn every key tool needed for AI success: Python, NumPy, Jupyter Notebooks, Pandas, Matplot-
lib, Sklearn, Keras and TensorFlow—all in one program.

What jobs will this course prepare me for?


This bootcamp is designed to build your skills in machine learning as well as in deep learning. As
such, it doesn't prepare you for a specific job, it will expands your knowledge and skills with artificial
intelligence. These skills can be applied to various applications and also qualify you to pursue
further studies in the AI-field.

How do I know if this program is right for me?


If you are interested in artificial intelligence and machine learning, this bootcamp is the perfect way
to get started and take your career to the next level!

Head to Head Support from AI Mentor:


Mentor will consistently remain with you throughout your bootcamp. As you build your AI skills and
working on your final project, mentors will provide head to head support and will always be available.

Software & Hardware:


Any 64-bit operating system with at least 4GB to 8GB RAM for this course. Furthermore, students
should install Python 3 and Jupyter Notebooks on their systems.

Languages:
English
Learning material and communication in English.

Introduction:
Kick start by learning the basics of python and applying them to create a simple linear regression
model using basic python. We will be setting up and working with Anaconda environment, as well as
work with jupyter notebook.

MACHINE LEARNERS
Machine Learning:
This module forms the basis of AI world. Machine learning brings together computer science and
statistics to build the predictive power. You will explore many different algorithms during this boot-
camp. It will include lots of preprocessing of data and its handling. You will learn to optimize your
models and tune your hyper-parameters to get the best result using the best features by utilizing the
tips on feature selection.

PROJECT

Machine Learning for Recommender systems


Recommender systems are one of the most successful and widespread application of machine learning
technologies in business. It can apply in scenarios where many users interact with many items.

Neural Networks
The new name of intelligent systems; Neural Nets. Learn the underlying concepts of this beautiful
technology that can easily understand the non linear patterns in data and produce perfect models to
make accurate predictions about future.

PROJECT

Solve Titanic Survival using Neural Networks


This will be a challenging task, that will be testing your preprocessing skills and all the ML skills that you
know. You will have to work hard to get an accuracy greater than 75%.

Computer Vision
Learn the basics of image processing and work with videos using the famous OpenCV library. Apply
different transforms and normalize the Image arrays. Prepare images and videos to be passed to
deep learning models to classify data in discrete sets.

Deep Learning
Hidden layers in a neural network is where the magic happens. Let’s convert this magic to science in
this module by learning the very simple concepts that let machines learn like humans. Acquire knowl-
edge of creating deep learning models.

MACHINE LEARNERS
Convolutional Neural Networks
Time to apply the skills learned from Computer Vision Module. Convolutional Neural Network are
special type of Nets that boosts AI tasks in Computer Vision domain. Construct knowledge base for
CNN, max pooling, batch nnormalization layers. Build models ready to be applied on real world
images.

PROJECT

Object Detection in Deep Learning


Object detection is one of the areas of computer vision that is maturing very rapidly.
Creating accurate Machine Learning Models which are capable of identifying and localizing multiple
objects in a single image remained a core challenge in computer vision. But, with recent advancements
in Deep Learning, Object Detection applications are easier to develop than ever before.

Natural Language Processing


Welcome to the other major field of AI. Let’s teach machine the way of understanding the natural
language like human do. Follow an intense pipeline for preparing raw texts ready for deep learning
models.

Recurrent Neural Networks


Prepare for Text Summaries and Time Series Data. Look at the evolution of Recurrent Neural Net-
works. Grasp the concepts of vanishing Gradient Descent and its solution. Practice RNNs by com-
pleting the project of the module.

PROJECT

Text Summarization Using Sentence Embeddings


Extract the most important information is the goal of text summarization, that uses a combination of
embedding and lstm layers.

And/Or

Image Captioning in Deep Learning


Caption generation is a challenging artificial intelligence problem where a textual description must be
generated for a given photograph.
Deep Learning Model to Automatically Describe Photographs, generating textual description of an
image.

MACHINE LEARNERS
Reinforcement Learning
Ever wonder how a child learn? Why not teach machine the same way. Take an intense tour of
Markov Decision process and Game theory to design a system that let machine realize its mistakes
and fix them by taking appropriate steps. Work in the field of robotics or create any other AI agent
you like.

PROJECT

Build a Game Bot


There are lots of gamers out there who want innovative things in game. We all know that Artificial intelli-
gence is not overtaking the tech world, but here we will build a game bot that will play as a human
alongside a player.
Bot will evolve the experience of gamers and it will start a new ear of gaming.
(Build Game Bot using OpenAI's Gym and Universe libraries)

Generative Adversarial Networks


Till now thing has been predicted not generated. GANs are special network that are able to generate
new patterns of data while looking at the old patterns. Develop an understanding of implementing
GANs to produce completely new images and sounds.

MACHINE LEARNERS
Course
Content

AI is everywhere, its surprising. Let's dive right in and think with AI. This module will help you to create an
understanding of machine learning algorithms and end-to-end process of investigating data through a
machine learning lens. Explore different types of ML problems like classification and regression. Learn to
identify patterns and underlying details in a given data.Cluster the chunks of data in specific groups.
Wander through valleys of linear and logistic regression producing some beautiful predictions using evil
datasets.

Welcome to the Program


Meet the course instructors and EML team.
Learn about the resources available to help you succeed.
Introduction to Evolve Machine Learners.

Historical Perspectives of Artificial Intelligence & Machine Learning


Historical context
Computer vision & AI in early eras
Types of Machine Learning

Introduction to Python
Learn basic of python: syntax, installation, conditions, loops, lists, functions, class.

Linear Algebra, Statistics and Optimization


Basic concepts and notation.
Learn vector and matrix notation.
Matrix multiplication.
Finding initial facts about data to choose an appropriate Algorithm.
Framework/Library: Numpy, Pandas.

Machine Learning Foundations


Machine Learning Libraries hand frameworks.
Data preproposing.
Basic pipeline of any Machine Learning.
Dataset:Boston Housing Prices.
Framework/Library: Numpy, Pandas, Sklearn, Matplotlib.

MACHINE LEARNERS
Logistic Regression
Build basic ML model and learn underlying algorithm.
Get to know the sigmoid function and its use.
Datasets:Iris , Titanic Survival Prediction.
Framework/Library: Numpy, Pandas, Sklearn, Matplotlib.

Machine Learning Methods


Learn by exploring core machine learning concepts, before moving on to supervised
and unsupervised learning.

Model Evaluation
Learning confusion matrix, accuracy and different metrics.

Support Vector Machine


Explore the dataset and prepare it for training and testing.
Learn the algorithm and its internal working along with its usage.
Datasets: Credit Card, Skin Segmentation Dataset.
Framework/Library: Numpy, Pandas, Sklearn, Matplotlib.

K means Clustering and K Nearest Neighbour


An introduction to unsupervised learning, use wine dataset to cluster different types of wines.
Datasets: Wine.
Framework/Library: Sklearn.

Decision Tree and Random Forest


Learn working with ensemble models and their benefits.
Learn the algorithm and its internal working along with its usage.
This dataset is generated using skin textures from face images of diversity of age, gender, and
race people. We will classify the skin intro different categories using decision trees.
Classify digits from 0 to 9 using a set of features with Random Forest Algorithm.
Datasets: Skin Segmentation Dataset, Pen-Based Recognition of Handwritten Digits Data Set.
Framework/Library: Numpy, Pandas, Sklearn, Matplotlib.

PROJECT

Machine Learning for Recommender systems


Recommender systems are one of the most successful and widespread application of machine learning
technologies in business. It can apply in scenarios where many users interact with many items.

MACHINE LEARNERS
See how humans has tried to replicate the human brain by making the state of the art neural
networks. Create non-linear models to capture the complex patterns in data. Look at various activa-
tion functions, their usages and drawbacks. Expand this understanding to move further to work with
images and sounds.

Introduction to Neural Networks


Learn about neural network
Learn solid foundations on deep learning and neural networks. You'll also implement gradient
descent and backpropagation in python.
Simple perceptron.
Multi-layer perceptrons.
Multi-layer feed-forward.
Get a hands-on introduction to TensorFlow, Google's deep learning framework.
Build first neural network.
Dataset: MNIST.
Framework/Library: TensorFlow,Numpy, Pandas.

Activation functions
Learn and compare different activation functions.
Step function, Sigmoid.
ReLU (Rectified Linear Unit), tanh, Softmax and Cross-entropy.
Why we use activation functions with neural networks?
How to choose activation function?
Implement Gradient Descent
How neural networks learns or improves?
Implementing the gradient descent from scratch using numpy matrix multiplication.
Optimization, stochastic gradient descent (SGD) and bash gradient descent (BGD).
Dataset: Movie reviews, MNIST.
Data Preprocessing: Feature Normalisation
Get a hands-on introduction to data labeling.
Learn more about data-preprocessing.
Data-post processing and Improving generalization.
Apply data scaling, remove outliers and get knowledge of feature selection.
Training an Artificial Neural Network
Learn several techniques to improve their training.
Finetune your model using appropriate activations and weights.
Get a hands-on introduction to Keras, another deep learning framework.
Datasets: Artificial Characters Data Set.
Framework/Library: Keras, TensorFlow, Numpy, Pandas.

MACHINE LEARNERS
Evaluating Performance and Hidden Layer
How hidden layers improve accuracy, purpose of hidden layers.
Learn different optimizer and loss functions.
Seq API , functional API.
Datasets: MNIST, notMNIST.
Framework/Library: Keras, TensorFlow, Numpy, Pandas.

Deep Neural Network


Learn how to improve deep neural networks by using different techniques such as regularization,
hyperparameter tuning and regularization techniques.

Improving Deep Neural Network


Using L1 and L2 regularization.

PROJECT

Solve Titanic Survival using Neural Networks


This will be a challenging task, that will be testing your preprocessing skills and all the ML skills that you
know. You will have to work hard to get an accuracy greater than 75%.

Let's continue the journey by performing intense operation on images and learn how to build and
exploring state-of-the-art CNN models. Let's dive deep into the world of deep learning. Shake hands
with Convolutional Neural Networks, the perfect solution to handle tasks like image classification,
object detection and many more like these. Learn the concepts behind image processing using deep
learning and implement them on real world problems.

Convolutional Neural Networks


How to build convolutional neural networks and apply it to image data.
Learn to implement the foundational layers of CNNs (pooling, convolutions) and to stack them
properly in a deep network to solve multi-class image classification problems.

Build a Convolutional Neural Network using TensorFlow


Walk through an example Convolutional.
Neural Network (CNN) in TensorFlow.
Implement a Convolutional Neural Networks using concepts from LeNet,dropout and regularization
on tensorflow. We will improve accuracy by using Image preprocessing techniques.
You'll study the line-by-line breakdown of the code.
Datasets: GTSRB.
Framework/Library: TensorFlow.

MACHINE LEARNERS
Transfer Learning
Learn how to use a pretrained network on a new problem with transfer learning.
Models: Inception v3, Alexnet.
Framework/Library: Keras, TensorFlow, Numpy, Pandas.

Deep Dive in Convolutional Neural Networks


Training Neural Networks (ConvNets, CNNs) and autoencoders for image classification use
dimensionality reduction and feature learning.
Use feature-based workflows for object detection, recognition, and tracking.
Learn some tricks to improve model.
Activation functions.
Initialization.
Dropout.
Augment your training data.
Batch normalization.
Deep Learning Models & Frameworks.
Models: COCO, Cifar10.
Framework/Library: Keras, TensorFlow, Numpy, Pandas.
Deep Learning Models & Frameworks
Get knowledge of Deep learning, discuss its theory & applications.
Remain Inflexible in Practice.
Learn Multimodal information processing and Challenges for Deep Learning.
Have look on Deep Convolutional Neural Networks.
Work with network models using TensorFlow and Keras.
Build Deep Learning Model - Multi-layer Perceptron (MLP).
Models: Inception-v3, AlexNet, VGG-16, and VGG-19.
Framework/Library: Keras, TensorFlow, Numpy, Pandas.
Market Trends of Artificial Intelligence
Companies are using AI and its sub-technologies to optimize real world problems as AI is an
umbrella of several interlinked technologies, these include Natural Language Processing (NLP),
machine learning, cognitive computing, neural network, computer vision, robotics and its related
technologies.
Case Study: Flight control system.
Models: Flight Simulator.

Detection and Segmentation


Semantic segmentation
Object detection
Instance segmentation

MACHINE LEARNERS
Self Driving Car
Learn to use Deep Learning, Computer Vision and Machine Learning techniques to simulate
a Self-Driving Car with python.
Batch Generator, Fit Generator, Defining Nvidia Model.
Model : Udacity Self Driving Car Simulator.

PROJECT

Object Detection in Deep Learning


Creating accurate Deep Learning Models which are capable of identifying and localizing multiple
objects in a single image remained a core challenge in computer vision. But, with recent advancements
in Deep Learning, Object Detection applications are easier to develop than ever before.

In this Section, enhance your existing machine learning and deep learning skills with the addition of
natural language processing, you will build your own recurrent networks and long short-term
memory networks with Keras and TensorFlow and perform sentiment analysis. Use recurrent
networks to recognize and act on sequences of inputs.
Build a deep neural network that functions as part of an end-to-end machine translation pipeline.
Your completed pipeline will accept English text as input and return the French translation.
These skills can be used in various applications such as part of speech tagging and machine transla-
tion, among others. You’ll develop the skills you need to start applying natural language processing
techniques to real-world challenges and Applications.

Natural Language Processing


Learn the main techniques used in natural language processing.
Walk through NLP model.
See how text gets processed in order to use it in models.

Text processing and Embedding


Learn techniques such as tokenization and lemmatization.
Use the different text model for word embeddings.
Quick review of Word2Vec and GLoVe.

Natural Language Modeling


We will see why data cleaning (normalizing text data) such as dealing with whitespaces,
grammatical mistakes, are so important.
One hot encoding vs word2vector, we will see how words can be transferred into vector space
and define relations with each other in a sentence.
Sentiment analysis using nltk.

MACHINE LEARNERS
Recurrent Nerual Networks
Introducing Recurrent Neural Networks (RNNs).
Concept of Sequence-to-Sequence Learning.
RNN theory, vanishing gradient & updates to RNN.
Datasets: Milk Production.
Frameworks: TensorFlow.
Long Short-Term Memory(LSTM) Fashion Network
Evolution of RNNs.
Learn how LSTM & GRU works.
Datasets: Stock Market Price.
Framework/Library: TensorFlow, Numpy, Pandas.

Quick implementation of RNN and LST


Overview of what students have learnt during the last 10 weeks.

AutoEncoders
Learn Encoding & Decoding.
Learn Autoencoders and use it data compression, image denoising, and dimensionality reduction.
Build autoencoders using TensorFlow.
Datasets: MNIST.
Frameworks: TensorFlow.

PROJECT

Text Summarization Using Sentence Embeddings


Extract the most important information is the goal of text summarization, that uses a combination of
embedding and lstm layers.

And/Or

Image Captioning in Deep Learning


Caption generation is a challenging artificial intelligence problem where a textual description must be
generated for a given photograph.
Deep Learning Model to Automatically Describe Photographs, generating textual description of an
image.

MACHINE LEARNERS
Let’s face some more AI, Deep reinforcement learning is one of AI’s hottest fields. If you’re excited to
dive right in and tinker with AI, then games are a great place to start. In this section, we will cover the
theory and recent advances in reinforcement learning as well as latest techniques used to create
artificially intelligent agents that can solve a variety of complex tasks, with applications ranging from
gaming to robotics. This section is designed to enhance existing machine learning and deep learning
skills with the addition of reinforcement learning theory and programming techniques.

Reinforcement Learning: An Introduction


A friendly introduction to reinforcement learning.
History of game programming.

Markov Decision Processes


Learn about policies and value functions.
Derive the Bellman equations.
Learn Markov Decision Processes to solve real-world problems.
Datasets : Frozen lake.
Frameworks: TensorFlow.

Reinforcement Learning
Implementation of iterative policy evaluation, policy improvement, policy iteration, and value iteration.
Environment : OpenAI Gym.
Frameworks: TensorFlow.

Q-Learning
Solve TAXI-V2.
Design algorithm to solve a problem from the research community.

Environment : OPENAI GYM’S ( TAXI-V2 ).


Frameworks: TensorFlow.

Deep Q-Learning
Learn value-based reinforcement learning methods to complexproblems using deep neural
networks.
Learn how to implement a Deep Q-Network (DQN), along with reply memory and Prioritized
Replay.
Environment : OpenAI Gym ( Pacman ).
Frameworks: TensorFlow.

MACHINE LEARNERS
PROJECT

Build a Game Bot


There are lots of gamers out there who want innovative things in game. We all know that Artificial intelli-
gence is not overtaking the tech world, but here we will build a game bot that will play as a human
alongside a player.
Bot will evolve the experience of gamers and it will start a new ear of gaming.
(Build Game Bot using OpenAI's Gym and Universe libraries)

Understand Generative Adversarial Networks. Apply what your brand new skills in the project
while implementing a GAN, GANs’ potential is huge, because it can learn to mimic any distribution
of data and can generate things.

Generative Adversarial Networks


Introduction to Generative Adversarial Network
Learn to understand and implement the model to simulate realistic images.
Build pair of multi-layer neural networks and make them compete against each other in
order to generate realistic images.
GAN — Why it is so hard to train.
Dataset: MNIST.
Frameworks: TensorFlow.

MACHINE LEARNERS

Das könnte Ihnen auch gefallen