Several years ago, I created a fully connected neural network from scratch in C as a learning exercise. I followed the first few chapters of Michael Nielson's book 'Neural Networks and Deep Learning' that I highly recommend.

The network was designed to train on the MNIST number dataset, which is a well-known dataset used in many machine learning examples. The goal is to identify hand-written digits as any number between 0 and 9. The final network performed quite well and achieved 97.14% accuracy on the test dataset. Not bad for a bit of matrix algebra wrapped up in some C code.

Anyways, ever since then I've had the idea to create a little browser widget to let people use the model I trained in an interactive way. Of course, I was beaten to the punch once, twice, and many more times I'm sure. But even still I wanted to see how well my model would perform at this task.

Homemode Fully Connected Neural Network

Before starting to work on the widget, I beefed up my neural network a little bit and was able to train one with a 98.2% accuracy on the test MNIST data. I then used a web framework called Svelte to create a drawing and predicting widget. Since my model is all simple linear algebra, exporting the weights from C and hard-coding them into Javascript was not too much work. Libraries like Math.js made it pretty easy to recreate everything. The final product is the widget you see below. It runs entirely client side in the browser using my trained neural network.

* does not run on iOS Safari, possible macOS Safari as well *

If you tried a few numbers, you probably noticed that the predictions can often be rather poor. I found that it has a really difficult time with '1's , '0's, and '9's. It was a bit disappointing, even with a 98.2% accuracy on test data, it still has a lot of trouble with new numbers. My guess is that, due to the fully-connectedness of the network, it has a difficult time generalizing new data. Like, if a '1' is off to the side or at a wrong angle that isn't present in the training data then it will predict incorrectly.

Keras/Tensorflow Convolutional Neural Network

Another type of network often used on the MNIST data is a convolutional network. I won't go into the topic here but this explanation was pretty helpful for my understanding. Convolutional networks work so well on MNIST that it's actually one of the 'getting started' examples for Keras.

I wanted to see how much the widget would improve with a convolutional network instead of my fully-connected version. So, I followed the Keras example and trained one in Python that reached an accuracy of 99.3% on the test data. Crucially though, I believe that it generalizes much better and is therefore more tolerant to digits that may not be presented in exactly the same way as in the training data. And the results definitely show, in my testing it seems to predict the correct digit much more often then my homemade model.

Again, the widget below is running entirely in the browser using the the Tensorflow.js library. Tenforflow.js allowed me to export the model from Python and import it directly into the Svelte widget.

* does not run on iOS Safari, possible macOS Safari as well *

Embedding Widgets

Because the widgets run entirely client-side, feel free to embed them anywhere on your own site using the code snippets below. They are web components that use a shadow DOM so should always look the same no matter where they are embedded. Kind of like frame, but for the modern age.

<script src="https://www.cluoma.com/js/mnist_widget.js"></script>
<div><mnist-checker-widget /></div>

<script src="https://www.cluoma.com/js/mnist_convolution_widget.js"></script>
<div><mnist-convolution-checker-widget /></div>

Full source code for this project is posted on my GitHub.

bittyhttp is a C library that I've been working on that aims to make building web services in C as easy as possible. Check out my last post for a better description and to see some examples.

One use-case that I see for bittyhttp is adding HTTP functionality to existing C applications. For example, if there is a long-running process somewhere on a server, it would be easy to expose some of its configuration over a web interface using bittyhttp. So to target this use-case there are a few changes to bittyhttp to make this particular situation easier.

Users can now start the server in its own thread. For example:

int
main(int argc, char **argv)
{
    bhttp_server *server = bhttp_server_new();
    /* setup server, add handlers, etc. */
    bhttp_server_start(server, 1);
   
    /* continue with normal application execution */

    bhttp_server_stop(server);
    bhttp_server_free(server);
    return 0;
}

This starts bittyhttp in a separate thread and returns immediately. So one could insert this at the appropriate place in an application to easily spin off an HTTP server. Then, when needed, a simple call to bhttp_server_stop and bhttp_server_free will shut down the server and free resources.

In this mode it is also possible to register new callback handlers while the server is running, adding even more flexibility. Overall, this is a small update but I think it makes it much easier to use and incorporate into other applications.

In other developments, I've also been experimenting with some sort of Lua interface to allow writing callbacks in Lua. I've actually used this in another project of mine but I'm not particularly happy with the implementation. So it's still squarely in the experimental category.

bittyhttp is distributed under GPLv3. However, if you're interested in incorporating it into your project, and need another license, feel free to reach out.

Thanks for reading!

bittyhttp is a new library that I've been working on that aims to make building web services in C as easy as possible. Microservices and HTTP APIs are very common these days and bittyhttp offers the ability to implement these in C without much hassle. It takes care of running the server so all the user needs to do is implement their callbacks.

When using bittyhttp, the user registers handlers to URLs with a callback function pointer. If an HTTP request is received that matches the handler URL, the callback is executed. Inside the callback, information about the HTTP request is exposed and allows the user to decide how they would like to handle the request. If no handler is found, bittyhttp defaults to acting like a standard webserver.

Check out the full project on GitHub.

A Quick Example

For instance, we can register a simple handler like this:

bhttp_add_simple_handler(server,
                            BHTTP_GET | BHTTP_POST,  // http methods
                            "/helloworld",           // pattern to match
                            helloworld_handler);     // callback function

And then implement whatever logic we need in the callback like this:

int helloworld_handler(bhttp_request *req, bhttp_response *res)
{
    /* business logic */
    bstr bs;
    bstr_init(&bs);
    bstr_append_printf(&bs, "<html><p>Hello, world! from URL: %s</p><p>%s</p><p>%s</p></html>",
                      bstr_cstring(&req->uri),
                      bstr_cstring(&req->uri_path),
                      bstr_cstring(&req->uri_query));
    bhttp_res_set_body_text(res, bstr_cstring(&bs));
    bstr_free_contents(&bs);
    
    /* add custom headers and response code */
    bhttp_res_add_header(res, "content-type", "text/html");
    res->response_code = BHTTP_200_OK;
    return 0;
}

Because bittyhttp uses a separate thread to handle each request, some care needs to be taken in the callbacks to prevent race conditions. Changing data that is not allocated inside of a callback will often require the use of mutexes or similar data structures. A database connection pool, for example, would need to be properly managed.

Use Cases

I see bittyhttp having 2 primary use cases. The first being to implement often-used API endpoints in C, for performance reasons. Perhaps authentication endpoints get hit a lot and you want to speed these up.

The second is adding HTTP support to an existing C application. Maybe you have an existing long-running application on a server somewhere and you want to expose some of its configurations via HTTP. In that case, bittyhttp would be an excellent option.

bittyhttp is GPLv3 licensed, and available on GitHub, but I would be open to relicensing it for specific cases. If you are interested in using it in your application, feel free to get in contact with me at [email protected].

squid poll

As a fun, little proof-of-concept, I created a site called squid poll. It's basically a Straw Poll clone that uses bittyhttp as its API backend. Feel free to visit the site to create your own poll, or fill out the one below: