Pioneering fundamental research into AI efficiency.

We believe that by increasing the information density of data we can help make AI more impactful and useful for everyone.

Explore research

Get a Sneak Peek at the Future of Data Curation

Our groundbreaking research into data selection, recognized at the ICLR conference, is the foundation for an upcoming product that will make AI significantly more accurate and cost-effective.

Learn more

“Fundamental research is at the core of what we do at Granica and we've combined that with our team's systems engineering expertise to build a platform that drives information density and efficiency, at cloud scale.”

Andrea Montanari, Chief Scientist at Granica

Featured Research

Scaling Training Data with Lossy Image Compression

LOSSY COMPRESSION

To capture the trade-off between model performance and the optimal storage of training data, we propose a `storage scaling law' that describes the joint evolution of test error with sample size and number of bits per image.

Read paper

Towards a statistical theory of data selection under weak supervision

INFORMATION DISTILLATION

Given a sample of size N, it is often useful to select a subsample of smaller size n<N to be used for statistical estimation or learning. Such a data selection step is useful to reduce the requirements of data labeling and the computational complexity of learning.

Read paper

Sampling, Diffusion, and Stochastic Localization

ALGORITHMS

Diffusions are a successful technique to sample from high-dimensional distributions that are not given explicitly but rather learnt from a collection of samples. We generalize the construction of stochastic localization processes.

Read paper

Compressing Tabular Data via Latent Variable Estimation (ICML 2023)

TABULAR COMPRESSION

Data used for analytics and machine learning often take the form of tables with categorical entries. We introduce a family of lossless compression algorithms for such data.

Read paper

Inline Data Detection in Large Data Streams

LOSSLESS DATA REDUCTION

We present a a novel approach to data processing and reduction method that involves receiving an input data stream and computing a set of features that are representative of or unique to the stream.

Read paper

Efficient Data Deduplication through Sketch Computation and Similarity Metrics

LOSSLESS DATA REDUCTION

The methods provide a more efficient and effective way of handling large data streams, which can be particularly beneficial in applications that require real-time data processing and reduction.

Read paper

Let's build the future of AI efficiency, together.

A diversity of products requires a diversity of perspectives. We're on a mission to build tools that are easy to use and impossible to ignore — and we want your unique voice.

Whether you're a researcher, a systems engineer, an ML engineer, or simply someone interested in a more efficient future for AI, reach out.

View careers