Bits & Bytes
A few interesting ML and AI related tidbits from around the web over the past week or so:
China is building a huge AI-focused research and business park.
- The state-backed $2.1 billion technology park is part of China’s wider initiative to position themselves at the forefront of emerging markets. The 55 hectare park is expected to house 400 businesses and generate nearly $8 billion a year.
Richmond-based AI startup, Notch, acquired by Capital One.
- Fifteen months after Capital One created an in-house “Center for Machine Learning,” the company has reportedly acquired Notch, a data engineering and machine learning consulting firm.
LG distributes in house AI development tools throughout company.
- A few weeks after LG introduced ThinQ, the umbrella brand for the company’s smart home products, the company has announced availability of an in-house deep learning platform, called DeepThinQ, which is meant to facilitate AI technology development across platforms within the company.
Google releases preemptible GPUs with a 50% discount incentive.
- Preemptible GPUs work well with large machine learning tasks and other batch computational jobs, and Google is making them cheaperfor customers.
CEVA unveils new family of AI-processors designed for deep learning at the edge.
- As I mentioned previously, I’ll be keeping an eye on AI acceleration hardware products this year. CEVA’s new line is one such offering.
A new derivative-free optimization toolbox for deep learning.
- ZOOpt/ZOOjl is an interesting toolset that allows for optimization across data that is inconsistent or non-continuous, where standard algorithms like gradient descent which require differentiability would fall short.
