denadai2 / compx

Analyze spatial and temporal separation using information geometry

Hi there 👋

git config --global user.name "Marco De Nadai"

I am a Research Scientist at Spotify, a world-wide audio streaming and media services company. I do research on Graph Neural Networks (GNNs) and sequential models to recommend personalized talk content (e.g. podcasts) to millions of users. Previously, I have been an Applied Scientist at Zalando, a world-leading e-commerce platform with more than 48 million customers. I built Machine Learning solutions to recommend outfits (i.e. sets of clothes that match well together) to customers. For example, if a customer finds a pair of jeans they like, our algorithms recommend matching items (e.g. shoes, sweater) to complete the look.

• 🌱 I’m also interested in Machine Learning and Computer Vision. Check my research in Google Scholar
• 😄 My GitHub profile is not representative of my code quality. My code is usually hidden from the public!
• ⚡ I have a website

💬 LinkedIn | Twitter

Recent Papers:

• NeurIPS 2021: Smoothing the disentangled latent style space for unsupervised image-to-image translation
• Scientific Reports: Living in a pandemic: adaptation of individual mobility and social activity in the US
• CVPR 2021: Efficient training of visual transformers with small-size datasets