• timhutton/twitter-archive-parser: Python code to parse a Twitter archive and output in various ways
• Find Fediverse accounts of your Twitter followings
• Stephen Knight on Sean Carroll, Colin Wright, and the binary of sex – Why Evolution Is True
• How To Be Good ianleslie.substack.com
• The 2022 Winners of the National Book Awards
  Discover the winners and a dozen finalists for one of America’s most prestigious literary awards.
• The FTX collapse: the latest revelations (part three)
  Bankruptcy filings, private messages, and leaked balance sheets reveal some of the chaos at FTX. Here’s what we know now.
• Too much efficiency makes everything worse: overfitting and the strong version of Goodhart’s law
  This blog is intended to be a place to share ideas and results that are too weird, incomplete, or off-topic to turn into an academic paper, but that I think may be important. Let me know what you think! Contact links to the left.
• Large scale autonomous driving data preparation with Dataflow ML | Google Cloud Blog
  Developing autonomous driving technology is a battle with data, both from a volume and data format point of view. Sources include point cloud 3D data obtained from LIDAR, video data obtained from multiple cameras, GPS position information, millimeter-wave radar, steering and various sensor information. Even in a busy city, less than 1% of the raw data contains useful information, so substantial effort needs to be put into filtering, interpreting, and combining the information to be used in a coherent manner. One way to avoid all this data cleansing effort is to run simulations in 3D virtual space, which can generate this data to verify many patterns at scale, but  raw data obtained from actual driving will continue to be essential to test against real-world behavior. Exploring this data is not easy; it is a huge challenge for engineers to comb through data and isolate specific scenes of interest. Imagine trying to find a scene with green traffic lights and a specific number of pedestrians. It’s like finding a particular meme in a pile of DVDs.

  In this blog, we will walk through how Dataflow ML can be used in autonomous driving development to find a scene of interest using a data set of images.

• Accelerate your models to production with Google Cloud and PyTorch | Google Open Source Blog

Digest powered by RSS Digest