Joseph Tassarotti

Title: Joseph Tassarotti: Innovator in Probabilistic Programming

Introduction

Joseph Tassarotti is a notable inventor based in Pittsburgh, PA. He has made significant contributions to the field of probabilistic programming, holding 2 patents that showcase his innovative approach to data-parallel inference methods.

Latest Patents

Tassarotti's latest patents include a groundbreaking invention titled "Data-parallel probabilistic inference." This invention relates to a probabilistic programming compiler that generates data-parallel inference code to sample from probability distributions in models provided to the compiler. The compiler utilizes a modular framework that allows for the addition and removal of inference algorithm information, enabling it to generate inference code that implements various available algorithms. The compiler's use of an intermediate representation (IR) symbolically represents features of probability distributions, facilitating efficient parallel processing of inference computations on highly data-parallel architectures, such as GPUs.

Another significant patent is the "Sparse and data-parallel inference method and system for the latent Dirichlet allocation model." This patent describes a data-parallel and sparse algorithm for topic modeling based on a highly parallel algorithm for a Greedy Gibbs sampler. The Greedy Gibbs sampler is a Markov-Chain Monte Carlo algorithm that estimates topics in an unsupervised manner by estimating the parameters of the Latent Dirichlet Allocation (LDA) model. This algorithm is designed to be implemented on highly-parallel architectures, such as GPUs, and is modified to take advantage of data sparsity while maintaining parallelism.

Career Highlights

Joseph Tassarotti is currently employed at Oracle International Corporation, where he continues to develop innovative solutions in the realm of probabilistic programming. His work has significantly advanced the capabilities of data-parallel inference methods.

Collaborations

Tassarotti has collaborated with notable colleagues, including Jean-Baptiste Tristan and Guy Lewis Steele, Jr. These collaborations have further enriched his contributions to the field.

Conclusion

Joseph Tassarotti is a distinguished inventor whose work in probabilistic programming has led to significant advancements in data-parallel inference methods. His innovative patents reflect his commitment to enhancing computational efficiency and effectiveness in this evolving field.