Machine Learning

The Hands-on Machine Learning course is developed by Attila Houtkooper and the Trifork Amsterdam Machine Learning team. The team are on the cutting-edge of machine learning techniques and have worked on an impressive variety of real-world projects over the last few years.

Now it's your turn to benefit from their wealth of knowledge and practical experience.

Delve into the detail on how to actually apply machine learning to your daily work as a developer, whilst having a confident grasp of the underlying theories.

 

Python Pre-training:

Python is now one of the most commonly used programming languages when implementing Machine Learning. Due to this, we’ve created an online crash course (in video format) for you to complete prior to the training. This will ensure you have the fundamentals under control so we can deep dive into Machine Learning without any limitations. 

 

Topics Overview for Day 1

Introduction to Machine Learning

  • We will go through the history of and major areas in Machine Learning so that you are able to explain many of the uses of Machine Learning.
    Python, Pandas, Numpy and SciKit Learn
    • We will ascert our common knowledge of these commonly used frameworks so that you are well-equipped to do the subsequent exercises.
      Regression
      • We will use regression to predict future values of a continuous measure such as the price of honey and see how we can predict how it will change.
        Classification
        • We will compare some classification algorithms to build a classifier on flower data. We look at the following algorithms: k-Nearest Neighbor, Decision Trees and Neural Networks.
          Unsupervised learning through Clustering and Principal Component Analysis
          • In discussing clustering we discover the usefulness of this technique in finding the structure in data without having any labels.
            Under- & overfitting
            • We will discuss the whys, dos and don’ts of splitting your data into training, test and validation sets and the use of cross-validation for comparing different model configurations.
              Data preprocessing and Embedding
              • Likely every learning model requires our data to be represented as numbers. That is more trivial when talking about a price level, and less when we want to include the description on the label in the data. We will go into representing text as vectors as well as scaling and transforming data to get the representation that is best for learning.

               

                Topics Overview for Day 2

                TensorFlow & Keras libraries
                • By working with TensorFlow we will get to know this modern framework built to create highly scalable computational graphs. We will be using Keras to build a neural network on top of this powerful engine as it provides an easy to use implementation of popular Neural Network architectures.
                  Convolutional Neural Networks
                  • We will be working with image data using Convolutional Neural Networks to show how we can use this powerful architecture to build a model that is independent of data variation.
                    Recurrent Neural Networks and Long-Short Term Memory
                    • We will use Long-Short Term Memory RNNs to work with variable-length datasets such as texts or event series.
                      Transfer Learning
                      • Why do all the training work ourselves when we can use pre-trained models? Transfer Learning is about using (parts of) pre-trained models to have a generic base to build a specialized model off of.
                        GPU servers on Amazon Web Services
                        • No Machine Learning professional can afford not to be aware of how to offload computationally intensive training to the cloud. We’ll dive into how to accomplish this using Amazon Web Services.
                          Guide on implementation within your organization
                          • We’ll take a deeper dive into what it takes to implement Machine Learning in your organization. Working with ML in actual work setting carries some important parts. Assessing data quality, getting the client to commit to the success of the project, choosing the right strategy, all help to execute a successful ML project.

                               

                                Prerequisites

                                • Software development experience, knowledge of Python and basic statistics would give you an advantage but is not required
                                • Developers are expected to bring their own laptop

                                Delivery:

                                • Instructor-led course given by skilled staff
                                • Best practices and know-how embedded in hands-on labs
                                • Interactive learning environment with strong Q&A
                                • Course delivered in English at our training rooms at 9 Rijnsburgstraat, Amsterdam

                                 

                                About the Trainer:

                                Attila is the CTO of Machine Learning at Trifork Amsterdam. He has worked in fintech, transport, healthcare and publishing. He facilitates the delivery of the best possible software by his team, by finding the optimal software quality-to-features ratio and striving for good software engineering, whilst ensuring delivery of the business requirements. Attila also has an MBA from the Erasmus University, which helps him to bring business principles to the software engineering world.

                                Upcoming training dates