Skip to content
Your progress

Checking your sign-in…

Who's learning?

Each profile keeps its own read lessons, badges, streak, and practice scores on this device.

Lesson 10 of 11

8 min

Text size

Eight lessons, one line each. Keep the table open while you answer.

Two weights, one line to cross, and what moved when the answer was wrong
IdeaThe line to keep
Holding a rule insteadOne guesser keeps the examples and one keeps a rule of numbers. Each answers the same stated day from what it kept.
The weighted sumEach input is scaled by its own weight. The scaled values add into one sum.
The line it crossesThe sum is compared to one threshold. At it or past it, the rule calls. Under it, it waits.
Measuring one wrongThe missed count needs its denominator. It comes from a real walk over the kept-back days.
Nudging the weightsOne learning step moves each weight. It moves by the nudge times its own input. The move goes toward the right answer.
Counting what movedGuess, compare, count, nudge, again. A count over a few days can hide what a nudge broke.
Write the sum yourselfThe weighted sum and the comparison are two short functions. At or past it says call. The day the sum equals the threshold says call.
What the word neural buysA neuron here is one unit. It adds inputs times weights, then compares the sum to a threshold. A network is many units joined. This track built one unit.

The pile and the rule answer the same day. What does each one keep?

  • Both keep a stored verdict for every day
  • The pile keeps examples; the rule keeps two weights and a threshold
  • The pile keeps weights; the rule keeps examples

The gauge weight is 6. What does that number do?

  • It records how many times the tank was filled this month
  • It tells the caretaker how full the tank is
  • It scales the gauge input: one step adds 6 to the sum

Gauge 5, days 3, weights 6 and 4. What is the weighted sum?

  • 42 — five sixes plus three fours
  • 8 — the inputs added, with no weights at all
  • 15 — the inputs multiplied together

The threshold is 40, and a day's sum is exactly 40. What does the rule say?

  • Wait — exactly on the line counts as under it
  • It depends on which weight is bigger
  • Call — at it or past it says call

In this track's learning step, what happens to the threshold?

  • It stays 40. Only the two weights move
  • It moves by the nudge too, in the same direction
  • It doubles whenever a weight moves up

The count is the rule missing 2 of the 4 kept-back days. Why say "of 4"?

  • The denominator is the count of weights the rule holds
  • Missed 2 of 4 and missed 2 of 40 are different claims
  • A count is only allowed when it comes from kept-back days

Gauge 6, days 0, and the rule waits when the caretaker called. Which weight moves?

  • Both weights, each by the full nudge of 1
  • The threshold, since the rule waited on a call day
  • Only the gauge weight, since the days input is 0

The rule says call and the caretaker's word is wait. Which way do the weights move?

  • Up, but by half as much
  • Down, each by the nudge times its own input
  • They do not move and the threshold moves instead

One nudge cuts the kept-back misses from 2 to 1. Misses on all 12 days go from 2 to 3. What does that show?

  • One nudge can fix a day and break others
  • The nudge was added up wrongly
  • Kept-back days should be counted twice

The log holds gauge 5, days 3 with the word call. The same numbers also carry the word wait. What follows?

  • A careful choice of weights can get both days right
  • The rule should drop both days from the log and carry on
  • Any rule of two weights must miss one of the two days

Why does "missed 0 of 1 day just learned" look too kind?

  • The count forgot to give its denominator
  • It measures only the day the step was fitted to
  • The step moved the threshold, so the count is off

In your weightedSum function, what should an empty inputs list return?

  • 0, since `sum` starts at 0 and the walk adds nothing
  • An error, since the walk has nothing to add up
  • The weights list, returned unchanged

Your decide function uses a strict "greater than". Which day catches it?

  • Every day, on the very first run
  • No day at all. Both comparisons always give the same answer
  • The hidden day whose sum equals the threshold exactly

In a textbook network drawing, which layer takes in the data?

  • The output layer
  • The hidden layer
  • The input layer

What is the honest, one-line description of what this term built?

  • One unit: inputs, weights, a sum, a threshold, a verdict
  • A neural network, joined from many units
  • A small working copy of a real brain cell, made much smaller

Assertion (A): Some two weights get every day of the log right. Reason (R): Two log days share inputs but carry different words. What is right?

  • A and R are true, and R explains A.
  • A is true, but R is false.
  • A is false, but R is true.

That is the whole track: two weights, one line to cross. And the nudge that moved them both.

You watched a rule change, and made it change yourself. Your next track is AI Ethics & Society. It asks who set the weights, and who a guess affects.

Reveal every answer to finish this lesson.