Machine learning (ML) and deep learning (DL) content from the past 24 hours. Plus, some AI stuff.
In an interview published on Forbes, SAS CEO Jim Goodnight notes that Viya, SAS’ new architecture, is the company’s third attempt (since 2012) at massively parallel computing, and “this time we’ve got everything figured out.” Customers who licensed software built on the first two architectures could not be reached for comment.
Gil Press offers twelve observations from the O’Reilly AI Conference.
Alex Woodie profiles RISELab, the next thing at Berkeley after AMPLab.
Methods and Techniques
— Adit Deshpande explains generative adversarial nets, a deep learning architecture that performs well for image classification.
— I covered this story yesterday, but it’s worth repeating: an MIT team develops a method to improve the interpretability of neural networks.
— On the Algorithmia blog, Stephanie Kim offers an introduction to ML for developers.
— Vincent Granville summarizes five case studies on designing better ML algorithms: click-fraud detection; ad matching; price optimization; detecting fake reviews and image recognition.
Software and Services
— In an article about Spark, Craig Stedman quotes some old guy.
— Jason Brownlee summarizes the growing use of Python for ML.
— On the Databricks blog, Tim Hunter et. al. explain GPU acceleration in Databricks.
— Boston-based DataRobot delivers a new release with TensorFlow support plus enhancements that improve interpretability and operational deployment.
— Mark Littlefield et. al. discuss the potential for ML/DL with high-performance embedded computing (HPEC).
— BabbyCam uses a crib camera, deep learning and a slew of NVIDIA Tesla GPUs to predict Sudden Unexpected Infant Deaths (SIDS).
— Joseph Lichterman describes how the Associated Press wants to use ML to turn print stories into broadcast stories.
— Horus, a device marketed by Swiss startup Eyra, uses stereo cameras and ML to assist the blind.