Machine Learning In-Depth
Details
Today we will dive deeper into the machine learning matter. Our first speaker will show you how to address advanced ML tasks with Amazon SageMaker.
Our second speaker will introduce her work on image retrieval, she will survey algorithms and give insights on different techniques and challenges.
You can also follow this session on our live stream: https://goo.gl/spf8sr
____
Title:
Advanced Machine Learning with Amazon SageMaker
Abstract:
Amazon SageMaker is a fully-managed platform that enables developers and data scientists to quickly and easily build, train, and deploy machine learning models at any scale. In this session, we’ll start with a quick overview of SageMaker. Then, using Jupyter notebooks, we’ll dive into the more advanced features of this service: state of the art built-in algorithms, pipe mode, batch transforms, hyper parameter optimization, Deep Learning.
Speaker:
Julien Simon
___
Title:
Image Retrieval with Convolutional Neural Networks: The Case of Landmark Recognition
Abstract:
Image recognition is the ability of a software to identify objects, places, and people in images. An Image retrieval task is slightly different: it requires to find similar images to a query image among an image dataset.
In the last decade, the field of machine learning has made tremendous progress on addressing these visual tasks.
In our talk, we will introduce, evaluate and rank the strength of different image retrieval algorithms, in the context of the landmark recognition challenge - the task of predicting landmark labels from images. We will start with the general overview of instance retrieval methods and focus on Convolutional Neural Network based methods as they provide the best performance.
We will systematically examine the design process of image retrieval systems and Deep Image Retrieval algorithms in particular, starting from the choice of the dataset and the preprocessing steps, followed by the selection of a backbone network and the design of a complementary component: the set of layers on top of the main network.
Finally, we will conclude with the discussion of outstanding issues in image recognition and retrieval.
Speaker:
Ekaterina Khramtsova, is a research technician at SnT, in the Sedan Lab. She concluded her Master Thesis at Futjisu AI Research in Paris.
____
Sponsor's Corner:
This meetup is supported by SnT and Amazon
SnT Research Seminar:
New Models of Financial Risk in FinTech: Impact for early stage financing
Speaker: Prof. Nir Vulkan, University of Oxford
Thursday, 27 September 2018 14:00 - 15:00
Location: https://goo.gl/DSoqGx
Amazon is hiring!
Data Scientist - Kindle Content: https://goo.gl/3UaQqp
Data Scientist - Kindle Content: https://goo.gl/6JZdN1
Please use Data Science Luxembourg or Dzidas Martinaitis for a reference in order to assure a support from Amazon in the future. If you have any questions regarding the roles fill free to email dzidoriu amazon.lu