StudyBrixFree worksheets, study guides and courses

Artificial intelligence · University

One gradient-descent parameter update

The gradient and learning rate belong to the specified loss and parameterization.

Make a worksheet

Method and conditions

The gradient and learning rate belong to the specified loss and parameterization.

wnew = w − η∂L/∂w

Worked examples

At w=-1, a loss gradient is -11. With learning rate 0.2, calculate one gradient-descent update to w. Round the final result to 4 decimal places if needed.

Subtract learning-rate times gradient: -1−(0.2)(-11).

1.2

At w=11, a loss gradient is 1. With learning rate 0.5, calculate one gradient-descent update to w. Round the final result to 4 decimal places if needed.

Subtract learning-rate times gradient: 11−(0.5)(1).

10.5

Try it

  1. At w=-8, a loss gradient is -8. With learning rate 0.4, calculate one gradient-descent update to w. Round the final result to 4 decimal places if needed.

  2. At w=17, a loss gradient is -1. With learning rate 0.6, calculate one gradient-descent update to w. Round the final result to 4 decimal places if needed.

Check your work

Answers and marking notes
  1. -4.8

    Subtract learning-rate times gradient: -8−(0.4)(-8).

  2. 17.6

    Subtract learning-rate times gradient: 17−(0.6)(-1).

Further reading and review references

These are reference links, not copied exercises, institutional endorsement or proof of independent review.

AI-assisted preview; independent educational review required. Level labels describe prerequisites, not accreditation or official curriculum alignment.