How many deep learning models do companies typically have in production? A lot fewer than you’d think. 84% of companies had five or fewer models in production. For many teams, this process is simply too hard or too costly. We recently surveyed more than 290 machine learning engineers and data scientists to find out how… Read More Companies Lack Resources to Get Deep Learning Models into Production [Survey]
If you work in the world of deep learning, odds are you know all about EfficientNets, a family of models developed by Google researchers which achieve better accuracy with much smaller models than previous convolutional neural networks (CNNs). They are, more specifically, an image classification architecture that have set new records for accuracy on the… Read More The Challenges of EfficientNets And the Way Forward
Neural Magic to Speak at and Sponsor ODSC EAST 2020 Neural Magic is excited to be participating in the Open Data Science Conference East, also known as ODSC EAST, this April. The conference, to be held virtually (previously in Boston) April 14th-17th, will feature people and companies who are working on the cutting edge of… Read More The Software GPU, Pruning for Success & More at ODSC East
Everything we know about memory requirements in machine learning may be wrong. Today, when data scientists process deep learning models using a “throughput computing” device like a GPU, TPU, or similar hardware accelerator, they’re likely faced with a decision to shrink their model or input size to fit within the device’s memory limitations. Training a… Read More Challenging Memory Requirements and Performance Standards in ML
We at Neural Magic have been working on solving complex machine learning problems for years, and together our team comprises several decades of experience in the deep learning space. Since we started Neural Magic, we have spoken to and learned from thousands of Data Scientists and Machine Learning Engineers at conferences, small groups, and one-on-one… Read More Contribute to Neural Magic’s First Annual State of Deep Learning Report
Simply defined, machine learning-based visual search uses images rather than text to deliver results. In retail, for example, consumers can snap a photo of a jacket or purse they see in the real world, and use these photos to locate and purchase the item online. According to a recent Forbes article, the popularity of visual… Read More How to Use Machine Learning in Visual Search for Retail
I would like to put forth an idea that might seem counterintuitive given present-day hype: that human-scale machine learning infrastructures must more closely map to the human brain. In more technical terms, this means that they should be based on modifications to existing commodity Von-Neumann architecture CPUs, rather than on today’s popular “brain-inspired” massively parallel… Read More Counterintuitive Lessons: How to Improve Machine Learning
What are you doing on March 19th of this year? We’ll be at SXSW in Austin, TX, and our very own CEO Nir Shavit will be giving a talk! If you’ll be at SXSW, we hope to see you at his talk. Big Brain Burnout: Nir’s Talk for SXSW Nir will be speaking on the… Read More Neural Magic is Coming to SXSW 2020: What’s Wrong with AI Computing?
Recently, we’ve had the honor of being named to two local lists that celebrate Boston-area companies that demonstrate high growth potential and innovative technology. Boston has always been a hotbed of innovation, with prestigious universities like MIT and Harvard situated here, and a long history of achievement in the sciences, medicine, and technology. It’s not… Read More Built in Boston & Bostinno Honor Neural Magic for Potential
Nir’s take on the future of machine learning—where it’s heading and where it should be heading—can be seen as contrary to the current, prevailing wisdom.