Alterna · learning material
08 Checkpoint: Neuronets
Check the concepts, the manual calculations, and the causal relationships.
What this material covers
- Eight reviews of the mechanism
- Research and explain
Content and key concepts
Progress is available with JavaScript. View is not equal to development.
Eight reviews of the mechanism
Threshold: 7 out of 8. without a timer. Balls are given for answers, not for visits. This is a local self-check; the decisions will be shown after dispatch.
1. In batch lights...
Created separately for each line
Common examples
Equal to Target labels
2. Increase of hidden width...
Changes actual matriices and parameter accounts
Just adding pictures.
Guarantees improvement on test
3. The Gradient on W has a shupe...
Like W
Always a scalar.
Always [batch, batch]
4. Same hidden units under symmetrical conditions ...
They'll have to be different.
Can get the same updates.
Never affects output.
5. Parameter in affine layer 2 × 3?
Your numerical response
6. Network parameters 2x3x2x1?
Your numerical response
7. Mean vector[1.2.3]?
Your numerical response
8. Second coordinates x+.5ReLU(x) for x=[-1.2]?
Your numerical response
The result will appear after all the answers have been sent.
Research and explain
Compare width/dept and init scal with a fixed budget. Then turn off the residual in a separate experience. Do not call the fat four XOR points a test of geneticisation; indicate where the webs have actually been updated.
First, shape the hypothesis, then change one parameter. Compare the numbers to the formula; state the extreme case and the limit of applicability. The free conclusion is not evaluated by a dummy automatic score.