Joint Meetup Sicara x Kili

👋 The Next Meetup of the 2021/2022 season will be on Wednesday, March 30th in a hybrid format with Kili Technology

As usual, there will be two talks on the program. Each talk will be followed by a question and answer session with the speaker.

🗓 The program:

Image Search Engine and its applications at Kili Technology by Raphael Segerie

For modern deep learning to work, large amounts of annotated data are required. Even using pre-trained models, fine-tuning a model requires specific training data for the desired use case. Kili is a French start-up that wants to facilitate data annotation to help companies build data from unstructured data (images, text, video, voice, etc.). During my internship, I worked as a machine learning engineer to study different machine learning technologies that would be beneficial to Kili to make data annotation more ergonomic, faster and safer. My work has led to proposals for new ML services focused on image and text modalities.
During this meetup, we will present 3 prototypes: a semantic image search engine, an outlier detection algorithm, a batch review algorithm for similar images.
We will discuss the mathematics of these algorithms as well as multiple use cases to annotate data faster and more securely.

Carbon footprint of data-centric methods by Mathieu Soul

What is the carbon footprint of training and serving an AI model? As the climate crisis worsens, and the usage of artificial intelligence increases, new research seeks to estimate the impact of deep learning models on the environment. During this MeetUp, we will first overview the main environmental impacts of the various stages of an AI project’s lifecycle (hardware, training, serving). Then, we will consider a specific computer vision project (inspired by my current project at Sicara), and attempt to find actionable ways to reduce its carbon footprint.