Gradient Descent Simply Explained (with Example)

 

So… I’ll try to explain here the concept of gradient descent as simple as possible in order to provide some insight of what’s happening from a mathematical perspective and why the formula works. I’ll try to keep it short and split this into 2 chapters: theory and example - take it as a ELI5 linear regression tutorial.

Feel free to skip the mathy stuff and jump directly to the example if you feel that it might be easier to understand.

Theory and Formula

For the sake of simplicity, we’ll work in the 1D space: we’ll optimize a function that has only one coefficient so it is easier to plot and comprehend. The function can look like this:

f(x)=w⋅x+2

where we have to determine the value of w

such that the function successfully matches / approximates a set of known points.

Since our interest is to find the best coefficient, we’ll consider w

as a variable in our formulas and while computing the derivatives; x will be treated as a constant. In other words, we don’t compute the derivative with respect to x

since we don’t want to find values for it - we already have a set of inputs for the function, we’re not allowed to change them.

To properly grasp the gradient descent, as an optimization method, you need to know the following mathematical fact:

• The derivative of a function is positive when the function increases and is negative when the function decreases.

And writing this mathematically…

ddwf(w)>0→f(w)↗

ddwf(w)<0→f(w)↙

This is happening because the derivative can be seen as the slope of a function’s plot at a given point. I won’t go into details here, but check out the graph below - it should help.

Why is this important?

Because, as you probably know already, gradient descent attempts to minimize the error function (aka cost function).

Now, assuming we use the MSE (Mean Squared Error) function, we have something that looks like this:

yi^=f(xi)

MSE=1n⋅∑i=1i=n(yi−yi^)2

Where: yi

is the correct value, yi^ is the current (computed) value and n is the number of points we’re using to compute the MSE

.

The MSE is always positive (since it’s a sum of squared values) and therefore has a known minimum, which is 0 - so it can be minimized using the aforementioned method.

Take a look at the plot below: the sign of the slope provides useful information of where the minimum of the function is. We can use the value of the slope (the derivative) to adjust the value of the coefficient w (i.e.: w = w - slope).

Time to compute the derivative. Before that, I must warn you: it’s quite a long formula but I tried to do it step by step. Behold!

ddwMSE=ddw(1n⋅∑i=1i=n(yi−yi^)2)=

=1n⋅ddw(∑i=1i=n(yi−yi^)2)=

=1n⋅∑i=1i=nddw((yi−yi^)2)=

=1n⋅∑i=1i=nddw((yi−yi^)2)=

=2n⋅∑i=1i=n(yi−yi^)⋅(−1)⋅dyi^dw

Phew. From here, you’d have to replace dyi^dw

with the derivative of the function you chose to optimize. For y^=w⋅x+2

, we get:

=2n⋅∑i=1i=n(yi−yi^)⋅(−1)⋅x

And that’s about it. You can now update the values of your coefficient w

using the following formula:

w=w−learning_rate⋅ddwMSE(w)

Example

We’ll do the example in a 2D space, in order to represent a basic linear regression (a Perceptron without an activation function). Given the function below:

f(x)=w1⋅x+w2

we have to find w1

and w2

, using gradient descent, so it approximates the following set of points:

f(1)=5,f(2)=7

We start by writing the MSE:

MSE=1n⋅∑i=1i=2(yi−(w1⋅x+w2))2

And then the differentiation part. Since there are 2 coefficients, we compute partial derivatives - each one corresponds to its coefficient.

For w1

:

∂∂w1(1n⋅∑i=1i=2(yi−(w1⋅xi+w2))2)=

=1n⋅∑i=1i=2(yi−∂∂w1(w1⋅xi+w2))2=

=1n⋅2⋅∑i=1i=2(yi−(w1⋅xi+w2))⋅(−1)⋅xi=

=−2n⋅∑i=1i=2(yi−(w1⋅xi+w2))⋅xi

For w2

:

∂∂w2(1n⋅∑i=1i=2(yi−(w1⋅xi+w2))2)=

=−2n⋅∑i=1i=2(yi−(w1⋅xi+w2))

Now, we pick some random values for our coefficients. Let’s say w1=9

and w2=10

.

We compute:

f(1)=9⋅1+10=19,f(2)=9⋅2+10=28

Obviously, these are not the outputs we’re looking for, so we’ll continue by adjusting the coefficients (we’ll consider a 0.15 learning rate):

w1=w1−learning_rate⋅∂∂w1MSE=

=9+0.15⋅22⋅∑i=1i=2(yi−(w1⋅xi+w2))⋅xi=

=9+0.15⋅((5−(9⋅1+10))⋅1+(7−(9⋅2+10))⋅2)=

=9−0.15⋅56=0.6

w2=w2−learning_rate⋅∂∂w2MSE=

=10+0.15⋅22⋅∑i=1i=2(yi−(w1⋅xi+w2))=

=10+0.15⋅((5−(9⋅1+10))+(7−(9⋅2+10)))=

=10−0.15⋅35=4.75

Recalculating the output of our function, we observe that the outputs are somehow closer to our expected values.

f(1)=0.6⋅1+4.75=5.35,f(2)=0.6⋅2+1.25=5.95

Running a second step of optimization:

w1=0.6+0.15⋅((5−(0.6⋅1+4.75))⋅1+(7−(0.6⋅2+4.75))⋅2)=

=0.6+0.15⋅1.75=0.86

w2=4.75+0.15⋅((5−(0.6⋅1+4.75))+(7−(0.6⋅2+4.75)))=

=4.75+0.15⋅0.7=4.85

Now, this is going to take multiple iterations in order to converge and we’re not going to do everything by hand. Writing this formula as a Python script yields the following results:

1: w1 = 9.000, w2 = 10.000, MSE: 318.5 
   f(1) = 19.000, f(2) = 28.000
------------------------------------------------
2: w1 = 0.600, w2 = 4.750, MSE: 0.6125 
   f(1) = 5.350, f(2) = 5.950
------------------------------------------------
3: w1 = 0.862, w2 = 4.855, MSE: 0.345603125 
   f(1) = 5.718, f(2) = 6.580
------------------------------------------------
4: w1 = 0.881, w2 = 4.810, MSE: 0.330451789063 
   f(1) = 5.691, f(2) = 6.572
------------------------------------------------
5: w1 = 0.906, w2 = 4.771, MSE: 0.316146225664 
   f(1) = 5.676, f(2) = 6.582
------------------------------------------------
6: w1 = 0.929, w2 = 4.732, MSE: 0.302460106908 
   f(1) = 5.662, f(2) = 6.591
------------------------------------------------
7: w1 = 0.953, w2 = 4.694, MSE: 0.289366466781 
   f(1) = 5.647, f(2) = 6.600
------------------------------------------------
8: w1 = 0.976, w2 = 4.657, MSE: 0.276839656487 
   f(1) = 5.633, f(2) = 6.609
------------------------------------------------
9: w1 = 0.998, w2 = 4.621, MSE: 0.264855137696 
   f(1) = 5.619, f(2) = 6.617


[...]


------------------------------------------------
195: w1 = 1.984, w2 = 3.026, MSE: 7.04866766459e-05 
     f(1) = 5.010, f(2) = 6.994
------------------------------------------------
196: w1 = 1.984, w2 = 3.026, MSE: 6.74352752985e-05 
     f(1) = 5.010, f(2) = 6.994
------------------------------------------------
197: w1 = 1.984, w2 = 3.025, MSE: 6.45159705491e-05 
     f(1) = 5.010, f(2) = 6.994
------------------------------------------------
198: w1 = 1.985, w2 = 3.025, MSE: 6.17230438739e-05 
     f(1) = 5.009, f(2) = 6.994
------------------------------------------------
199: w1 = 1.985, w2 = 3.024, MSE: 5.90510243065e-05 
     f(1) = 5.009, f(2) = 6.994
------------------------------------------------
200: w1 = 1.985, w2 = 3.024, MSE: 5.64946777215e-05 
     f(1) = 5.009, f(2) = 6.994

It converges to w1=2

and w2=3

which are, indeed, the coefficients we were looking for.

In practice, I recommend experimenting with smaller learning rates and more iterations - large learning rates can lead to divergence (the coefficients stray from their correct values and tend to plus or minus infinity).

Conclusion

I guess this is all. Reading it now, I think it might take more than 5 minutes but… I guess it’s still a short article when compared to others that discuss the same subject :))

I hope this proves useful as a starting point and you’ve got something out of it. Backpropagation of errors in nerual networks works in a similar fashion, although the number of dimensions is way larger than what was presented here. Aaand it contains some additional features in order to handle non-convex functions (and avoid getting stuck in local minima). Maybe in other article we’ll take a look at those, too.