In brief What’s the best way to chop up a banana so that each piece will perfectly fit onto a slice of bread that has been spread with peanut butter to make a classic sandwich?
An algorithm will do now it for you. Ethan Rosenthal, a data scientist at Square and presumed fan of delicious sandwiches, has described his model and provided step-by-step instructions so that you can train and build your own computer vision model for the optimal peanut butter and banana snack.
There are quite a few moving parts: You’ll need to take an image of said banana and bread, send the data to your model to analyze, plot the outputs, and then follow the instructions to slice the banana and stick them on specific locations on the surface of the bread. Given that each banana had various curvatures and thickness, the task isn’t as trivial as you might think in computational terms.
Follow the instructions here or just grab a knife and go to it yourself.
Google goes in the Navy
Google Cloud is going to help the US Navy maintain and protect its ships by analyzing images captured by drones with computer vision algorithms.
The Chocolate Factory won’t be dealing directly with the military. A company known as Simple Technology Solutions (STS), a software company based in Washington DC, will act as the middleman. STS was awarded the federal contract by the military’s Office of Naval Research (ONR) as part of its Phase I Small Business Innovation Research project.
It will use Google Cloud Auto ML, a suite of machine learning tools, to build a custom computer vision model to inspect drone footage of Navy vessels. The software will be trained to identify and, eventually, predict which areas are more likely to be damaged by corrosion.
“The manual inspection of Navy ships and vessels is a time-intensive, costly process that can drive up costs and slow down deployment," said Mike Daniels, vice president of Global Public Sector, Google Cloud.
"We're proud to work with the U.S. Navy and empower them with Google Cloud technology to transform corrosion inspections for greater efficiency and safety."
Google is reportedly going to start selling AI ethics advice by the end of this year.
A new AI residency programme at Apple
Apple has become the latest Silicon Valley giant to launch an AI residency program in a bit to attract more machine learning developers.
Being an AI engineer is lucrative; talent is in short supply and salaries are high. Residency programs allow companies to train tech-savvy graduates, who might then go to become future employees. Google, Facebook, Microsoft, and Nvidia have previously launched similar programs too.
“The year-long program will welcome residents with STEM graduate degrees or equivalent industry experience, software development backgrounds, and niche expertise — like design, linguistics, neuroscience, or psychology,” Apple said.
You don’t have to be an expert at AI algorithms already. The only main requirements are strong skills in Python, Objective-C, and C++. Positions are open in the UK, US, Switzerland, and Germany. You can apply here.
Graphcore’s next generation of AI chips will be 3nm
TSMC announced that it is working with British AI chip designer Graphcore to deliver 3nm node processors, according to Anandtech. The new chips are expected to arrive some time in 2022. TSMC is also working on pushing out 5nm silicon this year.
Graphcore’s second-generation Intelligence Processing Unit (IPU) known as the Colossus Mk2 GC200 was announced earlier this year, and is a 7nm TSMC-fabricated chip. ®