# backprop

Heterogeneous automatic differentation (backpropagation) https://github.com/mstksg/backprop#readme

Version on this page: | 0.0.3.0 |

LTS Haskell 10.5: | 0.0.3.0 |

Stackage Nightly 2018-02-18: | 0.1.3.0 |

Latest on Hackage: | 0.1.3.0 |

**Justin Le**

**justin@jle.im**

#### Module documentation for 0.0.3.0

# backprop

Automatic *heterogeneous* back-propagation.

Write your functions to compute your result, and the library will automatically generate functions to compute your gradient.

Differs from ad by offering full heterogeneity -- each intermediate step and the resulting value can have different types. Mostly intended for usage with gradient descent and other numeric optimization techniques.

Currently up on hackage (with 100% documentation coverage), but more up-to-date documentation is currently rendered on github pages!

## MNIST Digit Classifier Example

My blog post introduces the concepts in this library in the context of training a handwritten digit classifier. I recommend reading that first.

There are some literate haskell examples in the source, though (rendered as pdf here), which can be built (if stack is installed) using:

`$ ./Build.hs exe`

There is a follow-up tutorial on using the library with more advanced types, with extensible neural networks a la this blog post, available as literate haskell and also rendered as a PDF.

## Brief example

(This is a really brief version of my blog post)

The quick example below describes the running of a neural network with one
hidden layer to calculate its squared error with respect to target `targ`

,
which is parameterized by two weight matrices and two bias vectors.
Vector/matrix types are from the *hmatrix* package.

Let's make a data type to store our parameters, with convenient accessors using
*lens*:

```
data Network i h o = Net { _weight1 :: L h i
, _bias1 :: R h
, _weight2 :: L o h
, _bias2 :: R o
}
makeLenses ''Network
```

Normally, we might write code to "run" a neural network on an input like this:

```
neuralNet
:: R i
-> Network i h o
-> R h
neuralNet x n = z
where
y = logistic $ (n ^. weight1) #> x + (n ^. bias1)
z = logistic $ (n ^. weight2) #> y + (n ^. bias2)
logistic :: Floating a => a -> a
logistic x = 1 / (1 + exp (-x))
```

(`R i`

is an i-length vector, `L h i`

is an h-by-i matrix, etc., `#>`

is
matrix-vector multiplication, and `^.`

is access to a field via lens.)

When given an input vector and the network, we compute the result of the neural network ran on the input vector.

We can write it, instead, using *backprop*:

```
neuralNet
:: Reifies s W
=> BVar s (R i)
-> BVar s (Network i h o)
-> BVar s (R o)
neuralNet x n = z
where
y = logistic $ (n ^^. weight1) #> x + (n ^^. bias1)
z = logistic $ (n ^^. weight2) #> y + (n ^^. bias2)
logistic :: Floating a => a -> a
logistic x = 1 / (1 + exp (-x))
```

(`#>!`

is a backprop-aware version of `#>`

, and `^^.`

is access to a field via
lens in a `BVar`

)

And that's it! `neuralNet`

is now backpropagatable!

We can "run" it using `evalBP`

:

`evalBP (neuralNet (constVar x)) :: Network i h o -> R o`

And we can find the gradient using `gradBP`

:

`gradBP (neuralNet (constVar x)) :: Network i h o -> Network i h o`

If we write a function to compute errors:

```
netError
:: Reifies s W
=> BVar s (R i)
-> BVar s (R o)
-> BVar s (Network i h o)
-> BVar s Double
netError x targ n = norm_2 (neuralNet x - t)
```

(`norm_2`

is a backprop-aware euclidean norm)

Now, we can perform gradient descent!

```
gradDescent
:: R i
-> R o
-> Network i h o
-> Network i h o
gradDescent x targ n0 = n0 - 0.1 * gradient
where
gradient = gradBP (netError (constVar x) (constVar targ)) n0
```

Ta dah! We were able to compute the gradient of our error function, just by
only saying how to compute *the error itself*.

For a more fleshed out example, see my blog post and the MNIST tutorial (also rendered as a pdf)

## Lens Access

A lot of the friction of dealing with `BVar s a`

s instead of `a`

s directly is
alleviated with the lens interface.

With a lens, you can "view" and "set" items inside a `BVar`

, as if they were
the actual values:

```
(^.) :: a -> Lens' a b -> b
(^^.) :: BVar s a -> Lens' a b -> BVar s b
(.~) :: Lens' a b -> b -> a -> a
(.~~) :: Lens' a b -> BVar s b -> BVar s a -> BVar s a
```

And you can also extract multiple potential targets, as well, using
`Traversal`

s and `Prism`

s:

```
-- | Actually takes a Traversal, to be more general.
-- Can be used to implement "pattern matching" on BVars
(^?) :: a -> Prism' a b -> Maybe ( b)
(^^?) :: BVar s a -> Prism' a b -> Maybe (BVar s b)
(^..) :: a -> Traversal' a b -> [ b]
(^^..) :: BVar s a -> Traversal' a b -> [BVar s b]
```

Note that the library itself has no *lens* dependency, using *microlens*
instead.

## Benchmarks

Here are some basic benchmarks comparing the library's automatic differentiation process to "manual" differentiation by hand. When using the MNIST tutorial as an example:

For computing the gradient, there is about a 2.5ms overhead (or about 3.5x) compared to computing the gradients by hand. Some more profiling and investigation can be done, since there are two main sources of potential slow-downs:

- "Inefficient" gradient computations, because of automated differentiation not being as efficient as what you might get from doing things by hand and simplifying. This sort of cost is probably not avoidable.
- Overhead incurred by the book-keeping and actual automatic differentiating system, which involves keeping track of a dependency graph and propagating gradients backwards in memory. This sort of overhead is what we would be aiming to reduce.

It is unclear which one dominates the current slowdown.

However, it may be worth noting that this isn't necessarily a significant bottleneck.

*Updating*the networks using*hmatrix*actually dominates the runtime of the training. Manual gradient descent takes 3.2ms, so the extra overhead is about 60%-70%.Running the network (and the backprop-aware functions) incurs virtually zero overhead (about 4%), meaning that library authors could actually export backprop-aware functions by default and not lose any performance.

## Todo

Benchmark against competing back-propagation libraries like

*ad*, and auto-differentiating tensor libraries like*[grenade][]*Write tests!

Explore potentially ditching

`Num`

for another typeclass that only has`+`

,`0`

, and`1`

. Currently,`Num`

is required for all backpropagated types, but only`+`

,`fromInteger 0`

, and`fromInteger 1`

are ever used.The main upside to using

`Num`

is that it integrates well with the rest of the Haskell ecosystem, and many things already have useful`Num`

instances.There are two downsides -- one minor and one major.

It requires more work to make a type backpropagatable. Instead of writing only

`+`

,`0`

and`1`

, users must also define`*`

,`-`

or`negate`

,`abs`

,`signum`

, and all of`fromInteger`

. However, I don't see this being a big issue in practice, since most values that will be used with*backprop*would presumably also benefit from having a full`Num`

instance even without the need to backprop.Automatically generated prisms (used with

`^^?`

) work with tuples, and so cannot work out-of-the-box without a`Num`

instance for tuples. In addition, it's often useful to have anonymous products and tuples in general.This is bandaided-over by having

*backprop*provide canonical tuple-with-`Num`

types for different libraries to use, but it's not a perfect solution.This can be resolved by using the orphan instances in the

*[NumInstances][]*package. Still, there might be some headache for application developers if different libraries using*backprop*accidentally pull in their orphan instances from different places.Alternatively, one day we can get

`Num`

instances for tuples into*base*!

The extra complexity that would come from adding a custom typeclass just for

`+`

/`0`

/`1`

, though, I feel, might not be worth the benefit. The entire numeric Haskell ecosystem, at the time, revolves around`Num`

.However, it is worth noting that it wouldn't be too hard to add "Additive Typeclass" instances for any custom types -- one would just need to define

`(<+>) = (+)`

,`zero = fromInteger 0`

, and`one = fromInteger 1`

(a three-liner), so it might not be too bad.But really, a lot of this would all resolve itself if we got

`Num`

instances for tuples in base :)Explore opportunities for parallelization. There are some naive ways of directly parallelizing right now, but potential overhead should be investigated.

Some open questions:

a. Is it possible to support constructors with existential types?

## Changes

# Changelog

## Version 0.1.3.0

*Feb 12, 2018*

https://github.com/mstksg/backprop/releases/tag/v0.1.3.0

*Preulude.Backprop*module added with lifted versions of several*Prelude*and base functions.`liftOpX`

family of operators now have a more logical ordering for type variables. This change breaks backwards-compatibility.`opIsoN`

added to export list of*Numeric.Backprop*`noGrad`

and`noGrad1`

added to*Numeric.Backprop.Op*, for functions with no defined gradient.

*Internal*

Completely decoupled the internal implementation from

`Num`

, which showed some performance benefits.

## Version 0.1.2.0

*Feb 7, 2018*

https://github.com/mstksg/backprop/releases/tag/v0.1.2.0

- Added currying and uncurrying functions for tuples in
*Numeric.Backprop.Tuple*. `opIsoN`

, for isomorphisms between a tuple of values and a value.- (Internal) AD engine now using
`Any`

from*ghc-prim*instead of`Some I`

from*type-combinators*

## Version 0.1.1.0

*Feb 6, 2018*

https://github.com/mstksg/backprop/releases/tag/v0.1.1.0

- Added canonical strict tuple types with
`Num`

instances, in the module*Numeric.Backprop.Tuple*. This is meant to be a band-aid for the problem of orphan instances and potential mismatched tuple types. - Fixed bug in
`collectVar`

that occurs if container sizes change

*Internal*

Internal tweaks to the underlying automatic differentiation types that decouple backpropagation from

`Num`

, internally.`Num`

is now just used externally as a part of the API, which might someday be made optional.

## Version 0.1.0.0

*Feb 5, 2018*

https://github.com/mstksg/backprop/releases/tag/v0.1.0.0

- First non-alpha release.
More or less complete redesign of library. The entire API is completely changed, and there is no backwards compatibility!

- Everything is now "implicit" style, and there is no more
`BP`

monad. - Accessing items in
`BVar`

s is now lens-, prism-, and traversal- based, instead of iso- and generics-based. `Op`

is no longer monadic*Mono*modules are removed.*Implicit*modules are removed, since they are the default*Iso*module is removed, since`Iso`

s no longer play major role in the implementation of the library.

- Everything is now "implicit" style, and there is no more
- Removed dependency on
*ad*and*ad*-based ops, which had been pulling in the vast majority of dependencies. - Moved from
*.cabal*file to*hpack*system.

## Version 0.0.3.0

*Alpha*

https://github.com/mstksg/backprop/releases/tag/v0.0.3.0

Removed samples as registered executables in the cabal file, to reduce dependences to a bare minimum. For convenience, build script now also compiles the samples into the local directory if

*stack*is installed.Added experimental (unsafe) combinators for working with GADTs with existential types,

`withGADT`

, to*Numeric.Backprop*module.Fixed broken links in changelog.

## Version 0.0.2.0

*Alpha*

https://github.com/mstksg/backprop/releases/tag/v0.0.2.0

Added optimized numeric

`Op`

s, and re-write`Num`

/`Fractional`

/`Floating`

instances in terms of them.Removed all traces of

`Summer`

/`Unity`

from the library, eliminating a whole swath of "explicit-Summer"/"explicit-Unity" versions of functions. As a consequence, the library now only works with`Num`

instances. The API, however, is now much more simple.Benchmark suite added for MNIST example.

## Version 0.0.1.0

*Alpha*

https://github.com/mstksg/backprop/releases/tag/v0.0.1.0

Initial pre-release, as a request for comments. API is in a usable form and everything is fully documented, but there are definitely some things left to be done. (See [README.md][readme-0.0.1.0])