Deep Learning - Artificial Neural Networks with Tensorflow - Momentum

What analogy is used to explain the concept of momentum in gradient descent?

• A train moving on tracks

• A car accelerating on a highway

• A bird flying in the sky

• A ball rolling down a hill

How does gradient descent without momentum compare to pushing a box on sand?

• It moves faster with less effort

• It moves continuously without stopping

• It requires constant effort to keep moving

• It accelerates over time

In the mathematical formulation of momentum, what does the variable 'V' represent?

• Velocity

• Volume

• Voltage

• Viscosity

What is the typical value range for the momentum term 'MU'?

• 0.1 to 0.5

• 0.5 to 0.8

• 0.9 to 0.999

• 1.0 to 1.5

What is the main benefit of using momentum in optimization?

• It decreases the cost function

• It speeds up convergence

• It reduces the number of parameters

• It increases the learning rate

How does momentum help in scenarios with unequal gradients?

• By increasing the learning rate

• By reducing the learning rate

• By accumulating velocities in the shallow direction

• By ignoring the steep direction

What problem does momentum solve in optimization paths?

• It increases the cost function

• It reduces zigzagging and speeds up convergence

• It causes zigzagging across the valley

• It increases the number of iterations