Gating
AI we always wanted
What is Gating?
Shallow architecture
Here, neural networks rely on carry-over gates, which rewire the network for the subsequent inputs. With gates, artificial neural networks can be shallow. The networks grow laterally, not vertically. These networks can generate outputs only if a correct set of gates is activated. (*Patent pending)
Gates mimic similar mechanisms found in our brains
Gating is an invention based on insights on how human brain achieves intelligence. In the picture: metabotropic receptors and G protein-gated ion channels gate information flow in human brain.
An AI that repeatedly looks at its inputs
Two novelties
This drawing is from our provisional patent. Those who know how deep learning works will probably understand that this is something completely different, something They've never seen before. This is a revolutionary new way of making machines intelligent.
What we see are two key novelties. One is that the network is shallow, not deep. It expands laterally but not vertically. The second and most revolutionary novelty is that the network has gates that can open and close pathways, transiently re-wiring the network. The network inputs open and close gates, the network gets re-wired, and then a network has a new look on the inputs. After that, more gates are opened and closed, and the network again gets re-wired and has a new look into the input. And this repeats ad infinitum. The loop of re-wiring through gates and taking new looks never stops.
Helping machines becoming conscious
The organisation into Action and Gating regions is a gist of how our brain works, whereby a similar loop brings about the never-ending flow of our consciousness. Our cerebral cortex is the equivalent of a hugely expanded gating region (16 billion neurons to be precise, organized into 52 Brodmann areas).
Changing the world of AI
Gating is how AI will be working soon. We are about to replace deep learning with something much better, much more similar to how real brains work, something that can become conscious through the power of look-gate-rewire-look-again loop.
Our Vision:
AI we always wanted.
Our Mission:
We strive to develop AI products based on new machine learning technology — Gating — which solves many of the limitations of deep learning.
How does Gating accumulate knowledge?
Artificial Gated Neural Networks need to learn connections, and gates (among other things)
The learning process of a gated neural networks is more elaborate than that of deep learning. In addition to connections, also the properties of gates need to be adjusted during learning. In addition, the architecture of gated networks keeps expanding.
AI-Kindergarten
Training of gated networks requires AI-Kindergarten, which is an entire 'factory' for accumulation of knowledge. This includes learning curricula and many other components. Details of how AI-Kindergarten works are described in this document.
A theory behind Gating
The theory behind Gating and AI-Kindergarten is called practopoiesis.
Gating is driven by Danko Nikolić
- Neuroscientist (headed a lab at the Max-Planck Institute)
- AI aficionado (invented Guided Transfer Learning, co-organizer of the largest AI Meetup in Europe, co-author of a handbook on AI)
- Entrepreneur since the age of 18 (Companies co-founded: Digital Servis, ActiveSTIM, Robots Go Mental, TreviAI, Intelimensa).
Our science of Gating
Gating is a result of several decades of scientific work. Here are some key scientific publications that led to Gating technology.
- Nikolić, D. (2023). Where is the mind within the brain? Transient selection of subnetworks by metabotropic receptors and G protein-gated ion channels. Computational Biology and Chemistry, 103, 107820.
- Nikolić, D. (2015). Practopoiesis: Or how life fosters a mind. Journal of Theoretical Biology, 373, 40-61.
- Nikolić, D. (2017). Why deep neural nets cannot ever match biological intelligence and what to do about it?. International Journal of Automation and Computing, 14(5), 532-541.
- Lazar, A., Lewis, C., Fries, P., Singer, W., & Nikolić, D. (2021). Visual exposure enhances stimulus encoding and persistence in primary cortex. Proceedings of the National Academy of Sciences, 118(43), e2105276118.
- Nikolić, D., Häusler, S., Singer, W., & Maass, W. (2009). Distributed fading memory for stimulus properties in the primary visual cortex. PLoS Biology, 7(12), e1000260.
- Nikolić, D. (2009). Is synaesthesia actually ideaestesia? An inquiry into the nature of the phenomenon. In Proceedings of the Third International Congress on Synaesthesia, Science & Art (pp. 26-29).
Testimonials
Janet Winkler — CEO and co-founder bunton GmbH
Kevin Li — PhD Student at Harvard-MIT Health Sciences & Technology Division
Debjit D. Chaudhuri — Investor and Entrepreneur, Ex-CEO Infosys Germany
I have the pleasure of working with Danko on bunton projects, and I can confidently say that his mastery of AI and his entrepreneurial vision are unmatched. Danko is someone who not only understands the technical intricacies of AI but also has the rare ability to translate this knowledge into innovative, real-world solutions. His skil
I have the pleasure of working with Danko on bunton projects, and I can confidently say that his mastery of AI and his entrepreneurial vision are unmatched. Danko is someone who not only understands the technical intricacies of AI but also has the rare ability to translate this knowledge into innovative, real-world solutions. His skills are exceptional, driving success through a deep understanding of both technology and business. I highly recommend him as a true visionary in the field of AI and entrepreneurship.
Debjit D. Chaudhuri — Investor and Entrepreneur, Ex-CEO Infosys Germany
Kevin Li — PhD Student at Harvard-MIT Health Sciences & Technology Division
Debjit D. Chaudhuri — Investor and Entrepreneur, Ex-CEO Infosys Germany
I am confident that Danko is among the rare individuals capable of propelling humanity forward through AI. Achieving such a feat requires a profound understanding of AI and a keen sense of entrepreneurship—Danko embodies both.
My association with Danko began when he led AI development at a startup I invested in wherein, the startup quickly
I am confident that Danko is among the rare individuals capable of propelling humanity forward through AI. Achieving such a feat requires a profound understanding of AI and a keen sense of entrepreneurship—Danko embodies both.
My association with Danko began when he led AI development at a startup I invested in wherein, the startup quickly developed the leading AI solution for their use case, outperforming even established global corporations.
So, I was not surprised when Danko later unveiled his groundbreaking invention, Gating.
Working with Danko on various AI applications and assisting him in transforming his invention into a business has been a privilege.
Personally I like his positive attitude, integrity and professionalism as a wonderful team player.
Kevin Li — PhD Student at Harvard-MIT Health Sciences & Technology Division
Kevin Li — PhD Student at Harvard-MIT Health Sciences & Technology Division
Raul C. Muresan — CEO at Intelimensa & Neurodynamics, President at Transylvanian Institute of Neuroscience
A groundbreaking paradigm, Gating has the potential to revolutionize the field of artificial intelligence to an extent that hasn't been witnessed since the popularization of deep learning in 2012. With a strong theoretical foundation inspired by insight from rigorous neuroscience research, it holds the promise to address one of the most p
A groundbreaking paradigm, Gating has the potential to revolutionize the field of artificial intelligence to an extent that hasn't been witnessed since the popularization of deep learning in 2012. With a strong theoretical foundation inspired by insight from rigorous neuroscience research, it holds the promise to address one of the most pertinent challenges in modern AI: the challenge of creating a system capable of efficient and dynamic scaling to the complexity of the problem. By transferring the concept of "depth" in "deep learning" to the temporal axis, as in biological brains, Gating could allow for a level of generalizable learning yet to be imaginable for AI and could be a huge stride toward artificial general intelligence. I have always been mind-blown by the innovative originality of Danko's ideas and admired his courage to explore entirely novel paths in fundamental AI research, unafraid to build from the ground up if it meant making a game-changing development. Not only that, he has the ambition and leadership necessary to bring his idea into reality and introduce it to the world to transform others' work. His strong entrepreneurship as CEO of Robots Go Mental and his ability to clearly and enthusiastically illustrate his invention had led to my group at NASA Ames adapting Guided Transfer Learning, and I'm excited for the day that Gating comes to life and starts impacting projects around the world as well.
Raul C. Muresan — CEO at Intelimensa & Neurodynamics, President at Transylvanian Institute of Neuroscience
Raul C. Muresan — CEO at Intelimensa & Neurodynamics, President at Transylvanian Institute of Neuroscience
Raul C. Muresan — CEO at Intelimensa & Neurodynamics, President at Transylvanian Institute of Neuroscience
Gating will certainly leverage one of the most intriguing emerging
technologies with capabilities to imitate the human mind. I know Danko since over 20 years, both as a scientist and as an entrepreneur. He excels at both! His new technology is grounded in decades of intensive neuroscience research and reflects his relentless search for th
Gating will certainly leverage one of the most intriguing emerging
technologies with capabilities to imitate the human mind. I know Danko since over 20 years, both as a scientist and as an entrepreneur. He excels at both! His new technology is grounded in decades of intensive neuroscience research and reflects his relentless search for the neuronal mechanisms that support perception, cognition, and behavior. Gating is the result of Danko's incredible insights! I am sure it will usher a radically new and disruptive direction in AI. I very much look forward to see how this transformative technology will come to fruition in the next years.
Antonello Ceravola, Principal Scientist at Honda Research Institute Europe GmbH
Raul C. Muresan — CEO at Intelimensa & Neurodynamics, President at Transylvanian Institute of Neuroscience
Antonello Ceravola, Principal Scientist at Honda Research Institute Europe GmbH
I am excited about the Gating concept, which offers a transformative approach to AI learning—one that aligns more closely with the mechanisms of the human brain. Unlike traditional deep learning, this model promises learning from minimal data, pushing us towards the realm of a true one-shot learning. This is crucial, as it would allow AI
I am excited about the Gating concept, which offers a transformative approach to AI learning—one that aligns more closely with the mechanisms of the human brain. Unlike traditional deep learning, this model promises learning from minimal data, pushing us towards the realm of a true one-shot learning. This is crucial, as it would allow AI to adapt quickly and efficiently with limited computational resources, reflecting one of the ways we naturally learn from experience.
Moreover, the curriculum learning aspect brings a structured, progressive path to knowledge acquisition. Instead of being overwhelmed by vast, bulky datasets, this model promises to learn in stages, building upon previous experiences, making learning not only faster but more robust and meaningful. I believe this approach could be particularly powerful for AI systems that need to adapt continuously in dynamic environments, where real-time, incremental learning is essential.
Together, these attributes—learning with limited data and building upon a strong curriculum—I believe represent a significant leap forward for AI, enabling smarter, more adaptive, and more efficient systems.
Josko Jeraj, CEO TreviAI
Raul C. Muresan — CEO at Intelimensa & Neurodynamics, President at Transylvanian Institute of Neuroscience
Antonello Ceravola, Principal Scientist at Honda Research Institute Europe GmbH
I had the privilege of working closely with Danko Nikolic during his time as Head of AI and Data Science at our former company, evocenta. As the CEO of evocenta, I observed firsthand Danko’s remarkable technical expertise, leadership abilities, and entrepreneurial spirit.
Danko’s knowledge in AI and data science is truly exceptional. At ev
I had the privilege of working closely with Danko Nikolic during his time as Head of AI and Data Science at our former company, evocenta. As the CEO of evocenta, I observed firsthand Danko’s remarkable technical expertise, leadership abilities, and entrepreneurial spirit.
Danko’s knowledge in AI and data science is truly exceptional. At evocenta, he skillfully harnessed large language models to propel significant advancements in our projects, showcasing his deep understanding of complex algorithms and innovative problem-solving techniques. His talent for turning intricate technical concepts into practical, actionable strategies set new benchmarks for our team's success.
Beyond his technical prowess, Danko is a natural leader. He built our AI and Data Science team from scratch, bringing in top talent and cultivating a collaborative, high-performance environment. His leadership style is rooted in empathy, clear communication, and an unwavering commitment to excellence, which inspired the team to consistently deliver outstanding results.
Danko’s entrepreneurial mindset is another of his great strengths. He invented Guided Transfer Learning, a pioneering approach that greatly improves the efficiency and accuracy of machine learning models. This innovation led to the founding of a successful startup, Robots Go Mental, which has achieved significant milestones, including a collaboration with NASA to tackle complex real-world challenges.
Our professional journey continues today as we co-founded TreviAI, where Danko serves as Co-Founder and Chief Innovation and AI Officer (CIAO). His role is pivotal, leveraging his vast skill set and innovative technology stack to guide the company's vision and growth.
I am certain that Danko will bring the same level of dedication, creativity, and leadership to Gating. His combination of technical mastery, team-building skills, and entrepreneurial drive makes him exceptionally well-suited to lead Gating to new heights. I fully endorse Danko Nikolic and have every confidence that he will excel in his new venture.