Blog

AI for Social Good at NeurIPS

This year, the topic of AI for social good had a large presence at the 2018 Conference on Neural Information Processing Systems, better known as NeurIPS. XPrize, a non-profit organization with a mission to enable a better, safer, more sustainable world, kicked off the workshop with a presentation Sunday morning that provided an update on…

Read More

#NeurIPS

MLPerf Results Validate CPUs for Deep Learning Training

I have worked on optimizing and benchmarking computer performance for more than two decades, on platforms ranging from supercomputers and database servers to mobile devices. It is always fun to highlight performance results for the product you are building and compare them with others in the industry. SPEC*, LINPACK*, and TPC* have become familiar names…

Read More

Deep Defense: Training DNNs with Improved Adversarial Robustness

Though they are effective at a variety of computer vision tasks, deep neural networks (DNNs) have been shown to be vulnerable to attacks based on adversarial examples[1], or images perceptually similar to the real images but intentionally constructed to fool learning models. This has limited the application of DNNs in security-critical systems. My colleagues[2] and…

Read More

#NeurIPS

Accelerate AI with Intel® Architecture Powering Microsoft Azure

As AI continues its march into the mainstream enterprise, IT organizations are looking for ways to simplify implementation. Intel is helping customers accelerate their path to cloud-based AI by working with cloud providers to offer optimized and intelligent services based on trusted Intel® architecture. This week, the Microsoft Azure + AI Conference (co-located with DEVintersection)…

Read More

#Microsoft Azure + AI Conference

Precision and Recall for Time Series

The artificial intelligence (AI) revolution is having a profound impact on countless technologies. From automatic identification of our family and friends in social media to ordering up our next movie through voice commands, AI is everywhere. At Intel, we are dedicated to advancing AI across all domains, but we’re also working on research at the…

Read More

#NeurIPS

Scalable Methods for 8-bit Training of Neural Networks

Quantized neural networks (QNNs) are regularly used to improve network efficiency in deep learning. Though there has been much research into different quantization schemes, the number of bits required and the best quantization scheme is still unknown. In a paper jointly authored by myself and Itay Hubara, Elad Hoffer, and Daniel Soudry from The Technion…

Read More

#NeurIPS

HE-Transformer for nGraph

HE-Transformer for nGraph: Enabling Deep Learning on Encrypted Data

We are pleased to announce the open source release of HE-Transformer, a homomorphic encryption (HE) backend to nGraph, Intel's neural network compiler. HE allows computation on encrypted data. This capability, when applied to machine learning, allows data owners to gain valuable insights without exposing the underlying data; alternatively, it can enable model owners to protect…

Read More

#NeurIPS

Intel AI Research at NeurIPS 2018

Now that the excitement of the 32nd annual Conference on Neural Information Processing Systems (NeurIPS) is upon us, it’s a great time to take stock of everything that Intel AI research has worked on this year, and how much we’ve grown: we have more papers to present, more open source software tools to share, and…

Read More

#NeurIPS

Reinforcement Learning Coach + AWS SageMaker = Accelerated Deep Reinforcement Learning

At this year’s AWS re:Invent conference,  Amazon announced the integration of Intel’s Reinforcement Learning (RL) library, Reinforcement Learning Coach, with the Amazon SageMaker* platform for machine learning. SageMaker is an AI service that enables developers to build and train machine learning models for predictive or analytical applications in AWS’s cloud environment. Reinforcement Learning Coach is…

Read More

#AWS re:Invent#Reinforcement Learning Coach#Reinforcement Learning