Step by step instructions to Master Deep Learning, and Break into AI.
In the event that you need to break into AI, this Specialization will enable you to do as such. Deep Learning is a standout amongst the most very looked for after abilities in tech. This course will enable you to wind up plainly great at Deep Learning.
In five courses listed underneath, you will learn the foundations of Deep Learning, see how to build neural systems, and figure out how to lead successful machine learning projects. You will find out about Convolutional networks, RNNs, LSTM, Adam, Dropout, BatchNorm, Xavier/He initialization, and then some. You will chip away at case studies from medicinal services, autonomous driving, sign language reading, music generation, and natural language processing. You will ace the theory, as well as perceive how it is connected in industry. You will rehearse every one of these thoughts in Python and in TensorFlow, which we will educate.
AI is transforming multiple industries. In the wake of completing this specialization, you will probably discover inventive approaches to apply it to your work.
COURSE 1 : Neural Networks and Deep Learning
Commitment Needed: a month of study, 3-6 hours every week
About the Course
On the off chance that you need to break into cutting-edge AI, this course will enable you to do as such. Deep learning engineers are exceptionally looked for after, and acing Deep learning will give you various new vocation openings. Deep learning is additionally another “superpower” that will give you a chance to build AI frameworks that simply weren’t conceivable a couple of years back. In this course, you will take in the foundations of Deep learning. When you complete this class, you will: – Understand the major technology trends driving Deep Learning – Have the capacity to construct, prepare and apply completely associated Deep neural systems – Know how to implement effective (vectorized) neural systems – Understand the key parameters in a neural system’s architecture This course likewise shows you how Deep Learning really functions, as opposed to cursory or surface-level description. So in the wake of finishing it, you will have the capacity to apply Deep Learning out how to a your own particular applications. In the event that you are searching for an job in AI, after this course you will likewise have the capacity to answer fundamental interview questions.
Introduction to deep learning
Be able to explain the major trends driving the rise of deep learning, and understand where and how it is applied today.
Video · Welcome
Video · What is a neural network?
Video · Supervised Learning with Neural Networks
Video · Why is Deep Learning taking off?
Video · About this Course
Reading · Frequently Asked Questions
Video · Course Resources
Reading · How to use Discussion Forums
Quiz · Introduction to deep learning
Video · Geoffrey Hinton interview
Neural Networks Basics
Learn to set up a machine learning problem with a neural network mindset. Learn to use vectorization to speed up your models.
Video · Binary Classification
Video · Logistic Regression
Video · Logistic Regression Cost Function
Video · Gradient Descent
Video · Derivatives
Video · More Derivative Examples
Video · Computation graph
Video · Derivatives with a Computation Graph
Video · Logistic Regression Gradient Descent
Video · Gradient Descent on m Examples
Video · Vectorization
Video · More Examples of Vectorization
Video · Vectorizing Logistic Regression
Video · Vectorizing Logistic Regression’s Gradient Output
Video · Broadcasting in Python
Video · A note on python/numpy vectors
Video · Quick tour of Jupyter/iPython Notebooks
Video · Explanation of logistic regression cost function (optional)
Quiz · Neural Network Basics
Reading · Programming Assignment FAQ
Other · Python Basics with numpy (optional)
Practice Programming Assignment · Python Basics with numpy (optional)
Other · Logistic Regression with a Neural Network mindset
Programming Assignment · Logistic Regression with a Neural Network mindset
Video · Pieter Abbeel interview
Shallow neural networks
Learn to build a neural network with one hidden layer, using forward propagation and backpropagation.
Video · Neural Networks Overview
Video · Neural Network Representation
Video · Computing a Neural Network’s Output
Video · Vectorizing across multiple examples
Video · Explanation for Vectorized Implementation
Video · Activation functions
Video · Why do you need non-linear activation functions?
Video · Derivatives of activation functions
Video · Gradient descent for Neural Networks
Video · Backpropagation intuition (optional)
Video · Random Initialization
Quiz · Shallow Neural Networks
Other · Planar data classification with a hidden layer
Programming Assignment · Planar data classification with a hidden layer
Video · Ian Goodfellow interview
Deep Neural Networks
Understand the key computations underlying deep learning, use them to build and train deep neural networks, and apply it to computer vision.
Video · Deep L-layer neural network
Video · Forward and Backward Propogation
Video · Forward Propagation in a Deep Network
Video · Getting your matrix dimensions right
Video · Why deep representations?
Video · Building blocks of deep neural networks
Video · Parameters vs Hyperparameters
Video · What does this have to do with the brain?
Quiz · Key concepts on Deep Neural Networks
Other · Building your Deep Neural Network: Step by Step
Programming Assignment · Building your deep neural network: Step by Step
Other · Deep Neural Network – Application
Programming Assignment · Deep Neural Network Application
COURSE 2 : Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
- Commitment : 3 weeks, 3-6 hours per week
About the Course
This course will show you the “magic” of getting deep learning out how to function well. Instead of the Deep learning process being a discovery, you will understand what drives execution, and have the capacity to all the more deliberately get great outcomes. You will likewise learn TensorFlow. Following 3 weeks, you will: – Comprehend industry best-hones for building Deep learning applications. – Have the capacity to successfully utilize the basic neural system “traps”, including instatement, L2 and dropout regularization, Group standardization, inclination checking, – Have the capacity to implement and apply a variety of optimization algorithms, for example, such as mini-batch gradient descent, Momentum, RMSprop and Adam, and check for their convergence.. – See new prescribed procedures for the Deep learning period of how to set up prepare/dev/test sets and examine bias/variance – Have the capacity to implement a neural system in TensorFlow.
Practical aspects of Deep Learning
Video · Train / Dev / Test sets
Video · Bias / Variance
Video · Basic Recipe for Machine Learning
Video · Regularization
Video · Why regularization reduces overfitting?
Video · Dropout Regularization
Video · Understanding Dropout
Video · Other regularization methods
Video · Normalizing inputs
Video · Vanishing / Exploding gradients
Video · Weight Initialization for Deep Networks
Video · Numerical approximation of gradients
Video · Gradient checking
Video · Gradient Checking Implementation Notes
Quiz · Practical aspects of deep learning
Other · Initialization
Programming Assignment · Initialization
Other · Regularization
Programming Assignment · Regularization
Other · Gradient Checking
Programming Assignment · Gradient Checking
Video · Yoshua Bengio interview
Video · Mini-batch gradient descent
Video · Understanding mini-batch gradient descent
Video · Exponentially weighted averages
Video · Understanding exponentially weighted averages
Video · Bias correction in exponentially weighted averages
Video · Gradient descent with momentum
Video · RMSprop
Video · Adam optimization algorithm
Video · Learning rate decay
Video · The problem of local optima
Quiz · Optimization algorithms
Other · Optimization
Programming Assignment · Optimization
Video · Yuanqing Lin interview
Hyperparameter tuning, Batch Normalization and Programming Frameworks
Video · Tuning process
Video · Using an appropriate scale to pick hyperparameters
Video · Hyperparameters tuning in practice: Pandas vs. Caviar
Video · Normalizing activations in a network
Video · Fitting Batch Norm into a neural network
Video · Why does Batch Norm work?
Video · Batch Norm at test time
Video · Softmax Regression
Video · Training a softmax classifier
Video · Deep learning frameworks
Video · TensorFlow
Quiz · Hyperparameter tuning, Batch Normalization, Programming Frameworks
Other · Tensorflow
Programming Assignment · Tensorflow
COURSE 3 : Structuring Machine Learning Projects
- Commitment:2 weeks of study, 3-4 hours/week
About the Course
You will figure out how to build an effective machine learning project. On the off chance that you try to be a specialized pioneer in AI, and know how to set bearing for your collaboration, this course will demonstrate to you how. A lot of this substance has never been instructed somewhere else, and is drawn from experience building and dispatching numerous Deep learning products. This course additionally has two “flight simulators” that let you rehearse basic leadership as a machine learning project leader. This gives “industry encounter” that you may some way or another get simply following quite a while of ML work involvement. Following 2 weeks, you will: – See how to analyze mistakes in a machine learning framework, and – Have the capacity to organize the most encouraging headings for lessening blunder – Comprehend complex ML settings, for example, bungled preparing/test sets, and contrasting with as well as outperforming human-level execution – Know how to apply end-to-end learning, transfer learning, and multi-task learning.
ML Strategy (1)
Video · Why ML Strategy
Video · Orthogonalization
Video · Single number evaluation metric
Video · Satisficing and Optimizing metric
Video · Train/dev/test distributions
Video · Size of the dev and test sets
Video · When to change dev/test sets and metrics
Video · Why human-level performance?
Video · Avoidable bias
Video · Understanding human-level performance
Video · Surpassing human-level performance
Video · Improving your model performance
Reading · Machine Learning flight simulator
Quiz · Bird recognition in the city of Peacetopia (case study)
Video · Andrej Karpathy interview
ML Strategy (2)
Video · Carrying out error analysis
Video · Cleaning up incorrectly labeled data
Video · Build your first system quickly, then iterate
Video · Training and testing on different distributions
Video · Bias and Variance with mismatched data distributions
Video · Addressing data mismatch
Video · Transfer learning
Video · Multi-task learning
Video · What is end-to-end deep learning?
Video · Whether to use end-to-end deep learning
Quiz · Autonomous driving (case study)
Video · Ruslan Salakhutdinov interview
COURSE 4 : Convolutional Neural Networks
About the Course
This course will show you how to construct convolutional neural systems and apply it to image data. On account of Deep learning, computer vision is working obviously better than only two years back, and this is empowering various exciting applications going from safe self-ruling driving, to accurate face recognition, to automatic reading of radiology images. You will: – See how to construct a convolutional neural system, including recent variations such as residual networks. – Know how to apply convolutional systems to visual identification and acknowledgment undertakings. – Know to utilize neural style transfer to generate art. – Have the capacity to apply these algorithms to a variety of image, video, and other 2D or 3D data.
COURSE 5 : Sequence Models
About the Course
This course will show you how to build models for natural language, audio, and other sequence data. On account of Deep learning, sequence algorithms are working much better than only two years prior, and this is empowering various energizing applications in speech recognition, music synthesis, chatbots, machine translation, natural language understanding, and many others. You will: – See how to build and train Recurrent Neural Networks (RNNs), and commonly-used variants such as GRUs and LSTMs. – Be able to apply sequence models to natural language problems, including text synthesis. Have the capacity to apply sequence models to audio applications, including speech recognition and music synthesis.
Co-founder, Coursera; Adjunct Professor, Stanford University; formerly head of Baidu AI Group/Google Brain
Teaching Assistant – Younes Bensouda Mourri
Mathematical & Computational Sciences, Stanford University
Teaching Assistant – Kian Katanforoosh
M.S. Stanford University (Walter J. Gores 2017), B.S Ecole Centrale Paris
Details on the course can be found here.
Note: When these courses become available for free in future will update here.
PS: Financial Aid is available for learners who cannot afford the fee.