Visão Geral

Formação Machine Learning, O Machine Learning se concentra na criação de algoritmos para encontrar padrões ou fazer previsões a partir de dados experimentais. O crescente campo de aprendizado de máquina tem uma vasta gama de aplicações em diferentes áreas, como sistemas inteligentes, visão computacional, reconhecimento de fala, processamento de linguagem natural, robótica, finanças, recuperação de informações, saúde, previsão do tempo. O programa de Mestrado em Machine Learning desenvolve fundamentos teóricos e práticos necessários para estar na frente do progresso na próxima revolução técnica. Os aprimoramentos concluídos no Machine Learning e suas disciplinas relacionadas em breve rastrearão todas as partes da tecnologia.

Esta formação permite que você se torne especialista em várias abordagens de Machine Learning, como aprendizado supervisionado, aprendizado não supervisionado e processamento de linguagem natural. O programa de mestrado inclui treinamento sobre os recentes desenvolvimentos e métodos técnicos em Inteligência Artificial

Objetivo

Após realizar esta Formação Machine Learning, você será capaz de:

• Domine o conceito de programação Python.
• Entenda claramente o conceito de Machine Learning.
• Compreender o conceito de Deep Learning, Processamento de Linguagem Natural, Modelagem Gráfica e Aprendizagem por Reforço.
• Entenda a teoria subjacente aos algoritmos de aprendizado de máquina.
• Use o aprendizado de máquina para tomar decisões e previsões.
• Selecionar metodologias estatísticas e preditivas apropriadas.

Publico Alvo

• Engenheiros.
• Profissionais de software e TI.
• Profissionais de Dados.
• Cientistas de Dados.
• Profissionais de aprendizado de máquina.

Materiais

Conteúdo Programatico

Unit 1: Essential Mathematics - 20 hours

1. Include Linear Algebra which refers to familiarity with integrals, differentiation, differential equations, etc.
2. Statistics including Inferential Statics, Descriptive Statistics, Chi-Squared Tests, Random Variable, Gaussian and Normal Distributions, etc.
3. Probability like Bayes Theorem, Optimization like Convex Optimization, etc.

Unit 2: Introduction to Python - 20 hours

Mastering a programming language is highly necessary to pursue Data Science. Strongly recommend Python version 3

1. Python IDE
2. Understanding Operators
3. Variables and Data Types
4. Conditional Statements
5. Looping Constructs
6. Python Control structures
7. Function Modules
8. Functions
9. Data Structure
10. Lists
11. Dictionaries
12. Exception and file Handling
13. Handson Project

Unit 3: Python Libraries for Data Science - 10 hours

1. Understanding Standard Libraries and packages like Matplotlib
2. Numpy
3. Pandas & Scipy
4. Seaborn & Scikit-Learn,
5. BeautifulSoup, Bokeh, Urllib, etc.
6. Reading a CSV File in Python
7. Data Frames and basic operations with Data Frames
8. Indexing a Data Frame
9. Anaconda distribution

Unit 4: Common Machine Learning Algorithms and Intro to AI - 30 hours

1. Supervised, Unsupervised and Reinforcement Learning
2. Linear Regression,Logistic Regression,Decision Tree
3. K-Means,Random Forest
4. Seaborn & Scikit-Learn,
5. Dimensionality Reduction Algorithms - PCA
6. Gradient Boosting algorithms -GBM,XGBoost,LightGBM,CatBoost

Unit 5: Deep Learning - 20 hours

1. Artificial Neural Networks
2. Neurons, ANN & Working
3. Single Layer Perceptron Model
4. Multi-layer Neural Network
5. Cost Function Formation
6. Applying Gradient Descent Algorithm
7. Back-propagation Algorithm & Mathematical Modelling
8. Use Cases of ANN

Unit 6: Introduction to NLP- 20 hours

A) Text Preprocessing

1. Noise Removal
2. Lexicon Normalization
3. Lemmatization
4. Stemming
5. Object Standardization

B)Text to Features (Feature Engineering on text data)

1. Syntactic Parsing - Dependency Grammar,Part of Speech(POS) Tagging
2. Entity Parsing - Phrase Detection,Named Entity Recognition,Topic Modelling,N-Grams
3. Statistical features - TF – IDF,Frequency / Density Features,Readability Features
4. Word Embeddings

C)Important tasks of NLP

1. Text Classification
2. Text Matching -Levenshtein Distance,Phonetic Matching,Flexible String Matching
3. Co reference Resolution - document summarization, question answering, and information extraction
4. Other NLP problems / tasks - Text Summarization,Machine Translation ,NLG/NLU,OCR,Document to Information

D)Important NLP libraries

1. Scikit-learn: Machine learning in Python
2. Natural Language Toolkit (NLTK): The complete toolkit for all NLP techniques.
3. Pattern – A web mining module for the with tools for NLP and machine learning.
4. TextBlob – Easy to use NLP tools API, built on top of NLTK and Pattern.
5. spaCy – Industrial strength NLP with Python and Cython.
6. Gensim – Topic Modelling for Humans
7. Stanford Core NLP – NLP services and packages by Stanford NLP Group.

Unit 7: CNN and RNN - 20 hours

1. Convolutional Neural Networks (CNN)
2. Introduction to CNNs
3. CNNs Application
4. Architecture of a CNN
5. Convolution and Pooling layers in a CNN
6. Understanding and Visualizing a CNN
7. Image classification using Keras deep learning library
8. Recurrent Neural Networks (RNN)
9. Intro to RNN Model
10. Application use cases of RNN
11. Training RNNs with Backpropagation
12. Long Short-Term memory (LSTM)
13. Recurrent Neural Network Model
14. Deep Learning Frameworks

Unit 8: Deep Learning Frameworks and Tensorflow -10 hours

1. Introducing Tensors
2. Plane Vectors
3. Tensors
4. Installing TensorFlow
5. Getting Started With TensorFlow: Basics

To build a neural network and how to train, evaluate and optimize it with TensorFlow

1. TensorFlow Core: The main TensorFlow library which are widely popular for deep learning implementations
2.  Keras: Keras apis with TensorFlow backend only
3. TensorFlow Lite: Library for mobile/embedded device based lightweight solutions
4. TFX: TensorFlow Extended, a production scale platform for implementing end to end machine learning solutions. It is available on GitHub via 4 repos
5. TensorFlow Transform, TensorFlow Model Analysis, TensorFlow Serving, TensorFlow Data Validation

Unit 9: Capstone Project - 20 hours

1. Belgian Traffic Signs: Background/ MNIST/CIFAR Dataset
2. Loading And Exploring The Data
3. Traffic Sign Statistics
4. Visualizing The Traffic Signs
5. Feature Extraction
6. Re-scaling Images
7. Deep Learning With TensorFlow
8. Modeling The Neural Network
9. Running The Neural Network
10. Evaluating The Neural Network