The fastest training
Neural networks
on the market
Progress ANN – a new patented neural network architecture
Progress Inc.
Omega Neuron
Progress Inc. Omega Neuron
Progress ANN
Progress Artificial Neural Network
Progress
Progress
Progress
Progress
Progress
the Progress Neural Network
About PANN

The Progress Artificial Neural Network (PANN) is our new incomparably fast-trained artificial neural network with architecture closely emulating its biological prototype.

PANN Advantages
new artificial neuron
All classical neural networks are based on a 1943 'artificial neuron' design, roughly similar to the the nerve cells in our brain as they were imagined at the time. This design makes it possible to train such networks, but at the same time makes training a complex network take many months, if not years. The PANN technology is based on a new design of the artificial neuron (Omega neuron), more in line with modern biological knowledge. This provides training acceleration by hundreds and thousands of times, which opens up completely new possibilities for using neural networks.
High efficiency AI
PANN's training speed on CPUs and GPUs is a thousand times higher than that of existing ANNs. Based on available theoretical and experimental results, we can be sure that in the next 5 to 10 years it will be possible to create highly efficient artificial intelligence systems built on PANN.
Speed test
Caffe (CPU) 1x
Caffe (GPU) 11x
Caffe (cuDNN) 14x
PANN (CPU) 1,000x
NEW scale of efficiency
The PANN network can be implemented on a variety of hardware, ranging from a microchip or a video card to a smartphone, and up to a supercomputer. In any implementation, it will multiply the computing capabilities of the device, in fact, 'shifting the scale of efficiency'. This means that a laptop can work with the efficiency of a workstation, a workstation – with the efficiency of a mainframe, and a mainframe – with the efficiency of a supercomputer or a powerful computing center.
Extremely versatile​
PANN is a very versatile network capable of solving many problems that only specialized neural networks can solve today. It's also possible to build PANN based specialized networks that are much more efficient than existing ones. Furthermore, completely new types of neural network systems and commercial software products can be built using PANN to perform functions that require high system intelligence, and rapid training and retraining.
Details
Find out more

In classical neural networks, training is based on various versions of the ‘gradient descent’ method, widely used in mathematics, which requires a huge number of iterative computation cycles, while the training time grows exponentially with the amount of data used for training. In addition, one of the most unpleasant properties of such networks is that it’s almost impossible to ‘update’ their training or ‘retrain’ them – any change, no matter how small, in a system that has already been trained for a long time requires an equally long new training.

Because of all this, building and running such neural networks requires the use of very powerful computers for training, a lot of highly paid programmers, a lot of expensive computer time, and electricity. Consequently, this leads to very expensive and slow developments in the advent of improving such networks, due to the long testing times of experimental iterations. That’s exactly why such developments are currently almost completely monopolized by industry giants.

The Omega neuron made it possible to apply a fundamentally new training method that doesn’t require many iterations, providing an almost linear dependence of PANN’s training time on the amount of data. The more difficult the task and the larger the amount of data, the greater the gap in the learning rate compared to any other modern (convolutional, etc.) neural network. This will allow for the very rapid development of such networks and products based on them, both replacing existing ones and creating completely new ones.

Most people don’t understand what AI is and, just in case, they fear it, like they do everything unknown. But artificial intelligence is just the ability of artificial systems to perform some intellectual functions. Ancient animal traps were already the simplest AI devices. Today we are surrounded by a lot of smart devices – from the thermostat that regulates the temperature in the house to airplane autopilots, programs like Word or Excel, etc. However, it’s impossible to create a full-fledged AI with software alone. Those who program these systems cannot always foresee all the situations that may arise in real life. And classical neural networks are unsuitable for this task, since they cannot learn in real time.

We can expect the emergence of many PANN based ‘smart servants’ that will replace ‘hard programs’ or will control them in the interests of their ‘master’. These will be systems to which one could simply ‘explain’ what needs to be done, and they would determine how best to do it, how to do it within the law, how to ensure the safety of their owner and other people, etc.

In classical neural networks, training is based on various versions of the ‘gradient descent’ method, widely used in mathematics, which requires a huge number of iterative computation cycles, while the training time grows exponentially with the amount of data used for training. In addition, one of the most unpleasant properties of such networks is that it’s almost impossible to ‘update’ their training or ‘retrain’ them – any change, no matter how small, in a system that has already been trained for a long time requires an equally long new training.

Because of all this, building and running such neural networks requires the use of very powerful computers for training, a lot of highly paid programmers, a lot of expensive computer time, and electricity. Consequently, this leads to very expensive and slow developments in the advent of improving such networks, due to the long testing times of experimental iterations. That’s exactly why such developments are currently almost completely monopolized by industry giants.

The Omega neuron made it possible to apply a fundamentally new training method that doesn’t require many iterations, providing an almost linear dependence of PANN’s training time on the amount of data. The more difficult the task and the larger the amount of data, the greater the gap in the learning rate compared to any other modern (convolutional, etc.) neural network. This will allow for the very rapid development of such networks and products based on them, both replacing existing ones and creating completely new ones.

The first artificial neural network was the universal perceptron. It didn’t work very well for specific tasks, however. So for 60 years, development efforts went mainly into the creation of specialized neural networks ‘honed’ for solving specific intellectual problems.

There are many more or less narrowly specialized neural networks – cognitrons, neocognitrons, Hopfield, Kohonen, Kosko, Jordan Elman, Boltzmann networks, recurrent networks, resonance networks, deep learning networks, convolutional networks, etc. Bayesian networks of trust and networks of cellular automata are also close to them.

PANN allows us to reproduce the functions of any of these networks, all while doing it much more efficiently. PANN also allows to create new products, for example, highly efficient personal training systems, a ‘personal doctor’, or even an ‘Alter Ego’ (second self, the ideal secretary and agent on computer networks), etc.

Progress Inc.
Progress Inc.
Progress Inc.'s current IP portfolio
Patents across the world

We are constantly expanding our intellectual property portfolio by filing new patent applications.

Our Story
Progress journey
2012
Dmitri Pescianschi came up with the idea of a new type of artificial neuron that would allow to build neural networks with unique qualities, including super fast training.

2013

Dmitry shared his idea with his TRIZ teachers in the US, Boris Zlotin and Vladimir Proseanic. Together they founded Progress Inc. to further develop the new technology, test it, and patent it. First tests carried out.

2014

After running more tests and filing Provisional Patent Applications, work begins to protect Progress Inc.’s Intellectual Property with patents around the world

2015

Progress Inc. prepared and filed its first US patent application in September 2015. The same year we took part in the International AI Conference in Las Vegas.

2016

Progress Inc. received the first US patent for the PANN technology detailing the main ideas.

2017

Progress Inc. received the second US patent for the PANN technology detailing ways of implementing PANN on different platforms.

2018

First real-world test of the PANN technology through integration into a commercial product previously based on a standard neural network. Work begins on filing for patents in Japan, China, South Korea and other countries.

2019

Progress Inc. received the third US patent for the PANN technology detailing a new matrix training method for PANN networks.

2021

Progress Inc. is preparing to launch a benchmarking and experimentation platform to allow the general public to test PANN’s capabilities.

Established collaborations
Partners
Press coverage of Progress
Media About Us
Progress

…Lorem ipsum dolor sit amet, consectetur elit sed do eiusmod tempor incididunt ut labore et dolore magna aliqua enim

Progress

…Lorem ipsum dolor sit amet, consectetur elit sed do eiusmod tempor incididunt ut labore et dolore magna aliqua enim

Progress

…Lorem ipsum dolor sit amet, consectetur elit sed do eiusmod tempor incididunt ut labore et dolore magna aliqua enim

Scientific articles
Publications
Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
18 (Demo)
SIMPLE BLOG POST (Demo)
Lorem Ipsum. Proin gravida nibh vel velit auctor aliquet. Aenean sollicitudin, lorem quis bibendum auctor, nisi elit consequat ipsum, nec sagittis sem nibh id elit. Duis sed odio sit amet nibh vulputate cursus a sit amet mauris. Morbi accumsan ipsum velit.
19 (Demo)
SIMPLE BLOG POST (Demo)
Lorem Ipsum. Proin gravida nibh vel velit auctor aliquet. Aenean sollicitudin, lorem quis bibendum auctor, nisi elit consequat ipsum, nec sagittis sem nibh id elit. Duis sed odio sit amet nibh vulputate cursus a sit amet mauris. Morbi accumsan ipsum velit.
20 (Demo)
SIMPLE BLOG POST (Demo)
Lorem Ipsum. Proin gravida nibh vel velit auctor aliquet. Aenean sollicitudin, lorem quis bibendum auctor, nisi elit consequat ipsum, nec sagittis sem nibh id elit. Duis sed odio sit amet nibh vulputate cursus a sit amet mauris. Morbi accumsan ipsum velit.