docs: improve docs for Nx.conv/3 (#1564)

Co-authored-by: José Valim <[email protected]>

Author: polvalente

nx/guides/advanced/aggregation.livemd

@@ -93,7 +93,7 @@ m = ~MAT[
 >
 ```
 
-First, we'll compute the full-tensor aggregation. The calculations are developed below. We calculate an "array product" (aka [Hadamard product](<https://en.wikipedia.org/wiki/Hadamard_product_(matrices)#:~:text=In%20mathematics%2C%20the%20Hadamard%20product,elements%20i%2C%20j%20of%20the>), an element-wise product) of our tensor with the tensor of weights, then sum all the elements and divide by the sum of the weights.
+First, we'll compute the full-tensor aggregation. The calculations are developed below. We calculate an "array product" (aka [Hadamard product](https://en.wikipedia.org/wiki/Hadamard_product_(matrices)#:~:text=In%20mathematics%2C%20the%20Hadamard%20product,elements%20i%2C%20j%20of%20the), an element-wise product) of our tensor with the tensor of weights, then sum all the elements and divide by the sum of the weights.
 
 ```elixir
 w = ~MAT[

nx/lib/nx.ex

@@ -12919,6 +12919,11 @@ defmodule Nx do
   of summing the element-wise products in the window across
   each input channel.
 
+  > #### Kernel Reflection {: .info}
+  >
+  > See the note at the end of this section for more details
+  > on the convention for kernel reflection and conjugation.
+
   The ranks of both `input` and `kernel` must match. By
   default, both `input` and `kernel` are expected to have shapes
   of the following form:
@@ -13000,6 +13005,28 @@ defmodule Nx do
   in the same way as with `:feature_group_size`, however, the input
   tensor will be split into groups along the batch dimension.
 
+  > #### Convolution vs Correlation {: .tip}
+  >
+  > `conv/3` does not perform reversion nor conjugation of the kernel.
+  > This means that if you come from a Signal Processing background,
+  > you might call this operation "correlation" instead of convolution.
+  >
+  > If you need the proper Signal Processing convolution, you can use
+  > `reverse/2` and `conjugate/1`, like in the example:
+  >
+  > ```elixir
+  > axes = Nx.axes(kernel) |> Enum.drop(2)
+  >
+  > kernel =
+  >   if Nx.Type.complex?(Nx.type(kernel)) do
+  >     Nx.conjugate(Nx.reverse(kernel, axes: axes))
+  >   else
+  >     Nx.reverse(kernel, axes: axes)
+  >   end
+  >
+  > Nx.conv(img, kernel)
+  > ```
+
   ## Examples
 
       iex> left = Nx.iota({1, 1, 3, 3})

nx/lib/nx/defn/kernel.ex

@@ -1398,8 +1398,8 @@ defmodule Nx.Defn.Kernel do
   ## Named hooks
 
   It is possible to give names to the hooks. This allows them
-  to be defined or overridden by calling `Nx.Defn.jit/2` or
-  `Nx.Defn.stream/2`. Let's see an example:
+  to be defined or overridden by calling `Nx.Defn.jit/2`.
+  Let's see an example:
 
       defmodule Hooks do
         import Nx.Defn
@@ -1437,9 +1437,8 @@ defmodule Nx.Defn.Kernel do
         {add, mult}
       end
 
-  If a hook with the same name is given to `Nx.Defn.jit/2`
-  or `Nx.Defn.stream/2`, then it will override the default
-  callback.
+  If a hook with the same name is given to `Nx.Defn.jit/2`,
+  then it will override the default callback.
 
   ## Hooks and tokens