Check our new offer for Data Science Development Path!
Better image recognition for business
Deep learning is driving the development of computer vision, bringing greater speed and accuracy to tasks such as image recognition. On the strength of this growth, businesses across a range of sectors are putting image processing applications to work to improve existing processes and automate human-driven tasks.
Our training track will enable your team to build image recognition models more effectively.
The formula includes:
A FOUR-DAY HANDS-ON WORKSHOP
to transfer knowledge and work on real-life examples
A ONE-MONTH MENTORING PROGRAM
to help your team start their first project with our mentor’s support
Who it is for
Teams of up to eight engineers with no knowledge about machine learning.
Attendees should know at least one programming language. However, because the examples and projects will be done in Python, knowing at least the basics of Python syntax before starting the track is recommended.
Skills your team will gain
Knowledge of machine learning and deep learning concepts and algorithms that can be used for data analysis
Understanding image recognition using neural networks
The ability to use tools and metrics for evaluating deep learning models
The ability to create pipelines for solving real-life problems
Knowledge of how to manage data science projects and how it differs from managing software development projects
Four-day hands-on workshop
Day 1: Machine learning algorithms with Scikit-learn
Part 1: Introduction to machine learning
Data science processes
Supervised and unsupervised learning
Working with a dataset: terminology, splitting the dataset, normalizing data
Tuning the model and evaluating the results
Part 2: Hands-on coding
The programming environment: Python and its data science libraries, Jupyter Notebook
Data exploration and preprocessing
Using selected machine learning algorithms for classification and regression
Unsupervised learning: clustering
Day 3: Deep learning algorithms with Keras
Part 1: Introduction to Deep Learning
Building blocks for network structure
Training a neural network model
Layer types in detail: convolutional, max-pooling layers
Part 2: Hands-on coding
The programming environment: Keras, Neptune
Case study: fully connected network, convolutional neural network
Reducing the model’s overfitting
Day 2: Machine learning techniques
Part 1: Machine learning techniques + hands-on coding
Dimensionality reduction (PCA, t-SNE)
Random forest, XGBoost
Part 2: Managing the data science process – best practices
Version control: sources and datasets
Managing and tracking experiments
Day 4: Advanced deep learning technologies and brainstorming
Part 1: Hands-on coding – unboxing the black box
Neural network debugging
Visualizing what networks see: activations, saliency maps
Part 2: Highlight of advanced deep learning technologies
The goal of the mentoring program is to create an end-to-end model.
The mentoring program includes:
A dedicated mentor (an experienced data scientist)
Eight remote sessions (twice a week)
E-mail contact with a mentor in between the sessions
The final report – the summary of teamwork and recommendations for further team development
Additional materials (recommendations on the articles, books, blog posts, tools etc.)
Two remote mentoring sessions each week
Cleaning your dataset
Building reference solutions
Optimizing the model
Summary and conclusions
Because the mentoring will be more effective if you work on your use case, we encourage you to use your own image dataset. However, should you choose, we can provide an interesting and challenging case to be solved along with a dataset.
Let’s discuss your needs
Leave us a message and we will contact you shortly.