Вы находитесь на странице: 1из 2

2 days hands-on course on

Deep Learning: Convolutional


Neural Network with Python
(Keras and TensorFlow)
29 – 30 July 2019 School of Computer Sciences, USM
26 – 27 August 2019 Penang

Course description Objectives Fee


Deep learning is a powerful AI • To understand the concept & Students: RM 450.00
machine learning algorithm. It architecture of deep learning. Others : RM 600.00
uses convolutional neural • To understand the role of Group (more than 3
network (CNN) technology to do convolutional neural network participants): RM 500 pax.
machine learning tasks such as (CNN) in deep learning. Tea breaks, lunches,
object & event classification and • To design CNN model for workshop materials & cert.
recognition on a large images & solving problem. of attendance.
video dataset. It is known to be • To implement convolutional
the best tool for industrial 4.0 neural network model with Instructor
applications development. This Python, Keras and TensorFlow.
course teaches you how to • To evaluate and tuning the
construct, train, tune and CNN model for best result.
evaluate CNN model using
Phyton (TensorFlow and Keras) Who should attend
programming language.
The course is designed to suit Assoc. Prof. Dr. Putra Sumari
Topic covered beginner, student, researcher, Robotic, computer vision &
engineer, data analyst, software image processing group,
• Concept of deep learning and developer, product architect, School of Computer Sciences,
convolutional neural network. business owner, entrepreneur, USM, Penang.
• Nodes and connected layers, or any individual who wishes
weight & biases, feed forward to learn powerful deep learning
pass & back propagation, ReLU tool.
and activation function, stride &
padding, convolution, optimizer How to enroll
& reducing loss, max pooling,
metric and prediction. Online registration form at
• Programming with Python https://bit.ly/2LDM93v or scan
Tensor Flow, Keras, use cases. me.
Deep Learning: Convolutional Neural Network with
Python (Keras and TensorFlow)

DAY 1 BACKGROUND DAY 2: CONVOLUTIONAL NEURAL


NETWORK (CNN)
Topic 1 - Installing Tool for Deep
Learning Topic 5 - Use case: Fashion MNIST with
Neural Network (NN)
Installing Python & Jupyter Notebook
Installing TensorFlow, Keras and Matplotlib
Loading and exploring Fashion MNIST dataset
Setting up TensorFlow environment
Dealing with image
Topic 2 - Introduction to Setting training and testing dataset
Jupyter Notebook Developing model
Tuning epoch, learning rate, weight and biase
What is Jupyter Notebook Assessing performance metric and prediction
Jupyter Notebook and Jupyter Console
Exploring menu in Jupyter Notebook Topic 6 - Implementing CNN with
TensorFlow
Topic 3 - Programming with
TensorFlow
CNN vs NN: differences and similarities
Tensor Fashion MNIST use case
Constant, variable and placeholder Convolution layer
Graph and session Stride and padding
Tensorboard and visualization ReLU and pooling layer
Image operation Fully connected layer
Underfitting and overfitting
Topic 4 - Implementing Neural
Network (NN) with TensorFlow
Topic 7 - Use Case: Cat and Dog with CNN
Concept of deep learning
Concept and architecture of NN Loading and exploring cat and dog dataset
Artificial neuron & perceptron Setting training and testing dataset
Use case: Exclusive or (“XOR”) operation Creating placeholders and variables
Fully connected layer Creating convolutional layer
Activation function Creating ReLU layer
Feed forward pass: weight and biases Creating fully connected layer
Backpropagation: optimizer and reducing loss Setting cost function and optimizer
Learning rate, batch and epoch
Assessing performance
Performance metric and prediction

Start at 9 am and end at 5 pm. Enquiry: Ms Ramizah Ramino


Tea break at 10 am and 4 pm. ramizahramino@gmail.com
Lunch at 12.30 pm to 2.30 pm +60 (12) 495 2540

Вам также может понравиться