Matt is offline next week

[rhyolight]

I will be entirely offline next week chaperoning a science camp for my daughter’s 6th grade class. While I’m away, @Paul_Lamb will be acting as moderator of the forum. I’ll be back on Mar 18th and I’ll answer any messages or questions addressed to me.

[Bitking]

Searching through my collection …

Time to post the weird stuff while the boss is away!

[marty1885]

Fave fun!
And… time to go nuts. Ring the bell!

[Falco]

@rhyolight

I’ll be back on Mar 18th

Are you sure? Maybe you’ll need an extra week off to recover from all those little br***.
;-).

[Bitking]

Remember - science experiments that go boom are better!
Example:

[sheiser1]

Could I harass you to look at my update on this post first?

How much should I trust swarm results?

I think I may finally have a lead on this issue!! So here are the result objects for the first few time steps (resultObj = model.run(modelInput)). This model only has one field (the predicted field ‘media_wear_pct’) with min/max values around 600-100, though the predictions are coming back around ~250. It seems that the ‘bucketIndex’ is becoming the prediction in ‘multiStepPredictions’. Do you have any intuition how/why this might be? Maybe I haven’t properly engaged the classifier? Here is my …

[rhyolight]

I am PTO all next week again (family in town for Spring break, kids off school).

Sorry, no live-streams or forum questions answered. The @htm-cadre should be able to help folks out with any problems, or else contact @Paul_Lamb (forum mod) or @cmaver (if you need someone at Numenta).