Dimension reduction, State, Orientation, and Velocity.
Sent out the reasonably in-depth write-up of the proposed design for complex automatic clustering yesterday and expected to get at least a few questions or comments back; I ended up having to spend far more of my day than I wanted responding.
The good news is that the overall design is approved and one of our other lead MapReduce developers is up to speed on what we need to do. I’ll begin sending him some links and we’ll follow up on starting to generate code in between the sprints.
I haven’t gotten even a fraction of the time spent researching this that I wanted, so I’m well behind the learning curve as Phil blazes trails. My hope is that his lessons learned can help me come up to speed more quickly.
I’m going to continue some tutorials/videos today to get caught up so next week I can chase down the Protobuff format I need to generate with the comparison tensor.
I did get a chance to watch some more tutorials today covering cross-entropy loss method that made a lot of sense.
I went for a brief ride today (only 5 miles) and managed to fall off my bike for the first time. I went to stop at an intersection and unclipped my left foot fine, when I went to unclip my right foot, the cold-weather boot caught on the pedal and sent me crashing onto the curb. Fortunately I was bundled up enough that I didn’t get badly hurt, just bent my thumbnail back. Got back on the bike and completed the rest of the ride. I was still too sore to do the 12.5 mile today, especially in 20 mph winds.