EDITICON (1998)

In late 1990's I was experimenting with C/C++, web design, Javascript etc., years after I started programming for the first time in 1990-1991 in Basic on ZX Spectrum machines.

For the apps that I developed for MS-DOS in the '90s I needed icons for the buttons I added to my apps, so to help me out I coded EditIcon - a method of drawing images and shapes pixel by pixel and save them as .JCO files.

I came up back then with a novel extension .JCO (Jiby iCOn) because the files exported were somewhat .ICO files exported by the app EditIcon coded by Jiby (my nickname at that time for Nelu Lazar).

Over the years, I used the icons I designed and created with EditIcon for many apps that I developed in C/C++.

Basis for my neural networks class for letter recognition (1990s)

During those times at the end of the '90s and beginning of 2000s, a few students from a Computer Science college from Brașov, Romania came to me asking for my help to code for them a project they had to turn in for their lab or class. They showed me the class lecture and notes, showed me the requirements of the problem that they had to solve for that class, and asked me to help code the C++ class for them - because I was known in the area that I coded some cool stuff.

I wrote some details about this project on my website from those times - nelulazar.go.ro - and the page spotlighting the neural networks, machine learning code has been archived and preserved on Wayback Machine - Internet Archive website: https://web.archive.org/web/20020907145528/nelulazar.go.ro/Rn.htm

Here is how I translated, literal, my old Romanian blog about neural networks for text recognition into English:

Neural networks - Text recognition

"This application is based on the neural networks theory, having specific the possibility of text recognition. This program/app is not as complex as Recognita, for instance, but is a starting point toward solving problems such as text recognition. The user is able to draw, with the mouse, in a matrix, a shape of a letter, not necessarily looking alike the prototype letters of the app. Using methods related to the theory of implementing neural networks, the app will establish the potential similarity of the drawn letter, patchy, to one of the alphabet letters, of which prototypes were implemented and passed through the learning procedure. In such, in the right side of the screen the user will be able to visualize, in real time, the letter that they try to shape. In the top banner of the app there are the prototypes of the letters from the Romanian alphabet. The code uses classes for the implementation fo the graphical interface, as well as for the processing of the letters' recognition (the neural class and their functions). The app was made for a college project."

And my original blog post in Romanian about neural networks for text recognition:

Retele neuronale - Recunoasterea textelor

"Aceasta aplicatie se bazeaza pe teoria retelelor neuronale, specifica avand posibilitatea recunoasterii de texte. Acest program nu este unul de complexitatea lui Recognita, de exemplu, dar este un pas de pornire in vederea rezolvarii unor astfel de probleme, cum ar fi recunoasterile de text. Utilizatorul va putea introduce, cu ajutorul mouse-ului, intr-o matrice, o forma de litera, nu neaparat asemanatoare cu prototipurile aplicatiei. Prin metode ce apartin teoriei implementarii retelelor neuronale, programul va stabili asemanarea literei desenate, neuniforma, cu una dintre literele alfabetului, ale caror prototipuri le are implementate si trecute prin procedura de invatare. Astfel, in dreapta ecranului utilizatorul va putea sa vizualizeze, in mod real, litera careia incearca sa-i dea forma. In bara de sus a aplicatiei se afla prototipurile literelor alfabetului romanesc. Programul foloseste clase pentru implementarea interfetei grafice, cat si pentru realizarea recunoasterii literelor (clasa neuronilor si a functiilor acestora). Aplicatia a fost realizata pentru un proiect de facultate."

Snapshot of my neural networks text recognition blog post

Here is a snapshot for the archive of my blog post about neural networks for text recognition, as preserved by Wayback Machine - Internet Archive website: https://web.archive.org/web/20020907145528/nelulazar.go.ro/Rn.htm

Snapshot of EditIcon app emulated in DOSBox in 2026

This snapshot is of EditIcon's original .EXE compiled in the '90s running now in DOSBox, but it was briefly redacted to remove my personal information I added for some odd reason while coding it back in 1998.