Projects

These projects trace my research career from the MIT Media Lab through Microsoft Research and Telefónica Research. The sections below are organized by career period, with research areas retained within each period. More recent work is represented in the major ELLIS Alicante projects and my publications.

Major ELLIS Alicante projects

The following projects represent major recent work led or co-led by Nuria Oliver through ELLIS Alicante.

Beautyverse Research on beauty filters, aesthetic norms, diversity, and algorithmic bias, with work published in NeurIPS, CHI, and Royal Society Open Science. View project at ELLIS Alicante Trustworthy AI A sociotechnical research program connecting algorithmic fairness, human-AI collaboration, and AI regulation, including ICLR workshop work and a doctoral thesis. View project at ELLIS Alicante Human-Centric Federated Learning Privacy-preserving machine learning research addressing membership inference, participation, and the protection of personal data. View project at ELLIS Alicante Art Censorship on Social Media Research on algorithmic content moderation, artistic freedom, and the visibility of art in social platforms, including CHI 2024 work. View project at ELLIS Alicante Cognitive Biases and AI Research connecting human decision patterns with bias in language models and human-AI interaction. View project at ELLIS Alicante Maike: a Socratic Chatbot An open-source educational AI designed to support critical thinking and learner agency, with AIED work and classroom validation. View project at ELLIS Alicante XPRIZE Pandemic Response Challenge The winning $500,000 Cognizant-sponsored challenge team applying data science to pandemic response. View project at ELLIS Alicante

Telefónica Research

Projects from 2007–2016

Recommendation Systems

Research on recommender systems and the reliability of user ratings as preference signals.

Bicing: Shared Bicycle Behavior

Uses shared-bicycle data to uncover and predict urban mobility behavior.

MAMI Mobile Annotations

A mobile-phone prototype for annotating and searching digital photos with speech.

ThisOrThat Social Shopping

A social-shopping mobile application that gathers trusted friends’ opinions in context.

RelAltTab Window Switching

An enhanced Alt-Tab interface that helps users switch to related windows.

Microsoft Research

Projects from 2000–2007

TripleBeat Mobile Training

A mobile personal trainer combining music, competition, and physiological feedback.

SWISH Window Semantics

Semantic analysis of window titles and switching history to support task management.

HealthGear Mobile Sleep Apnea

A mobile wearable system for monitoring and analyzing sleep-apnea signals.

MPTrain Mobile Personal Trainer

A mobile personal trainer combining physiology, music, and exercise feedback.

S-Seer Perceptual Sensing

Uses value-of-information policies to guide perceptual sensing and analysis.

MIHMMs: Mutual Information Hidden Markov Models

Introduces mutual-information hidden Markov models for richer sequence modeling.

GWindows Stereo Interaction

Lightweight stereo vision for fluid, non-desktop computer interaction.

Seer Office Activity

Layered probabilistic representations for learning office activity from multiple streams.

MIT Media Lab

Projects from 1995–2001

Curve Analogies

Learns statistical models of two-dimensional curves for designing new shapes.

Image Analogies

Learns image filters by example to create new visual transformations automatically.

Driver Behavior Recognition

A SmartCar testbed and graphical-model framework for recognizing driver behavior.

An augmented-reality system that attaches video and audio experiences to real objects.

Wearable ASL Recognition

A wearable computer-vision system for recognizing American Sign Language.

Human Interaction Modeling

A real-time computer-vision system for modeling and recognizing human interactions.

Coupled HMMs for T'ai Chi

Coupled hidden Markov models for recognizing and classifying two-handed actions.

Reinforcement Learning in a Maze

Compares reinforcement-learning algorithms in a configurable maze-world environment.

Responsive Portraits

Interactive portraits that respond to a viewer’s movement and expression.

LAFTER Face Tracking

Real-time tracking of lips and faces with facial-expression recognition.

Audience Group Behavior

A multi-agent system for modeling audience behavior and group dynamics.