Secure Machine Learning

Subscribe • Previous Issues

Data Pegacorns

With Kenn So of Shasta Ventures, we identify data startups with real market traction. We chose companies founded 2006 or after as it coincides with the initial release of Hadoop, the open-source technology platform that started the big data era. This is the second in a series of posts on pegacorns, private companies that have at least $100M in annual revenue.

Read The Post

Data Exchange podcast

  • Narrative AIHilary Mason, co-founder of Hidden Door explains how they use machine learning and language models to empower and augment creators of role playing games.
  • Adversarial Machine LearningEdmon Begoli, leads the AI Systems R&D section at Oak Ridge National Laboratory (ORNL), where he is also a distinguished member of the ORNL research staff. 
  • Orchestrating Machine Learning Applications: Haytham Abuelfutuh is co-founder and CTO of Union, a startup founded by the team behind Flyte, a popular open source orchestration framework originated by Lyft.

Secure Multi-Party Computation and ML

A comprehensive data privacy and security policy involves protecting the confidentiality and integrity of data in any of these three states: at rest, in use, and in transit. Securing data while in use is of growing importance to data and AI teams tasked with building models that rely on sensitive data. The issue has been that tools have traditionally been too complex and/or too slow to be of practical use.  A new open source library addresses problems that have plagued solutions in this space.

A Secure Multi-Party Computation protocol (SMPC) allows a program to be executed by participants so that the output is revealed only to the desired parties, and that no inputs belonging to other parties will be revealed to participants other than what can be inferred from the outputs.

My friends at CipherMode Labs recently introduced algorithms and an architecture to process encrypted data that makes SMPC practical and easy to use. CipherCore is an exciting new open source, high performance library that makes SMPC accessible to data teams. It’s aimed at users and teams already familiar with tools like TensorFlow, PyTorch, and JAX.

Learn More

The Business Intelligence Index

Business intelligence (BI) tools enable non-programmers to transform data into insights that can be used to inform strategic and tactical business decisions. Modern BI tools let analysts visually analyze and interact with large data sets, and some tools incorporate data preparation and advanced analytics and modeling capabilities. We recently measured the popularity of existing BI tools, using an index that relies on public data and is modeled after TIOBE’s programming language index.

Read The Post

If you enjoyed this newsletter please support our work by encouraging your friends and colleagues to subscribe: