Dora Sabov

Background

About Dora Sabov

I grew up counting everything: my steps, the boxes on the right side of the road, the letters in every word. Since then, I knew I wanted to spread the word about mathematics and of course programming. In 2020 I have been working on a scientific work, which is called “Time series forecasting with a combination of methods”. You may ask, what is this project about? Let me tell you! An important step in forecasting is to check the predictions, i.e. to assess their accuracy and validity. The most influential models of the time series are defined in the learning phase of the combined model. In the learning process, the predictive model automatically adapts to the time series examined and eventually becomes a convex linear combination of the most influential basic models. This approach provides an opportunity to build an effective forecasting model that can be successfully used to forecast different time series in the fields of economics, medicine, and the social sphere. In this year, 2021, with my professor we decided to work on "Neural elements with discrete activation functions above the Galois field and their application”. We studied the algebraic and logical properties of multivalent neural elements over the Galois field and, based on the obtained results, developed a new spectral method for the synthesis of these elements. The specific examples we examined show that the functional capabilities of neural elements above the Galois field significantly exceed all known neural elements with discrete activation functions. This means that all the reflections that can be realized by these logical functions can be realized by a single neural element over the corresponding Galois field, and there is no need to build a neural network, as is the case with classical neural elements. As you can see, I really like math and diving deep into different projects. My others passion is statistics and Python.