SerriAI is our proprietary Discriminator- Generator(DG) model inspired by Generative Adversarial Networks (GANs). Including popular models, Neurales also gives the user access to our DG-model for their applications. Unlike many popular models, SerriAI does not need nearly as many learning samples to train, is non- iterative, and uses only CPU resources. It is optimized for and excels over current models in low resource environments with low data samples, low processing speeds, and low power.
Download Our Technical Paper Here
Our technology focuses on innovation in the three major pillars of AI:
After continually running the algorithm on a variety of real world examples, ranging form XOR gate to practical problems within major industries, SerriAI can easily achieve a accuracy of over 90% and up to 99% in certain tasks.
Once data is input into SerriAI, the training of the neural network takes seconds. The algorithm is conjectured to have linear running time.
With SerriAI there is no need for extensive computing power or GPUs to train the model. One of the first iterations of SerriAI was ran in Microsoft Excel on your typical laptop.