A significant performance gap exists between biological neurons and the artificial neurons used to power machine learning systems. 


Our patented solution, Perforated Backpropagation™, brings artificial neurons closer to the accuracy and efficiency of biological neurons.  

EARLY RESULTS

       +41% 

Accuracy Improvement on Pharmaceutical Dataset


+8%

Precision Score Improvement on Stock Forecasting Dataset

-89.3%


Reduction in Network Size on Patient Mortality Dataset while maintaining accuracy



HOW IT WORKS

ARTIFICIAL NEURONS

Artificial neurons, created in 1958, were designed based on our understanding of human neurons at the time. Despite significant advancements in neuroscience, this simplified model is still the one used in today's Deep Learning systems.

BIOLOGICAL NEURONS

Biological neurons are much more complex. They use their branches, known as dendrites, to interpret and filter the signals they receive. Dendrites play a crucial role in how neurons process information, making the human brain incredibly efficient and capable of complex thought.

OUR 

SOLUTION

Our patented technology brings the concept of dendrites into artificial neurons, making them more accurate and efficient. This update helps make computer models a bit more like how our brains actually work.

CURRENT STATUS