Assume that a new neural network procedure (NNNP) is found that allows to efficiently train neural networks up to the best their theoretical performance. Suppose that the NNNP is "miraculous" and can fulfill the expectations of the big players in the AI field.
But the mammoth companies are unreachable, as proven by the fact that they never answer, and maybe nobody in those corporations ever reads, emails with proposals about the NNNP. Thus, the corporations potentially more interested in NNNP are and will remain enclosed in their security rooms, unaware of the breakthrough.
Furthermore, the NNNP could have been found by someone living in a third world country, impoverished and without local access to venture capital. The following question arises:
HOW CAN SUCH NNNP BE MARKETED?
One obvious answer is: MOVE OUT OF THE THIRD WORLD. But there should be alternatives that do not require becoming an expatriate.
Another could be: START YOUR OWN BUSINESS, wherever you are. But that would convert a researcher into an entrepreneur with no time available for further basic development of new ideas.
Again, here is the question for discussion:
HOW CAN SUCH NNNP BE MARKETED?
"One obvious answer is: MOVE OUT OF THE THIRD WORLD. But there should be alternatives that do not require becoming an expatriate." some VCs would question this. you need to move to where the appropriate platform, system or centre is, that can permit the idea. cant really land a boeing in the desert. academic entrepreneurship comes to mind. patent the model, and write a paper on it. prove the concept and market it that way. or, find an industry contact, and build relationship. i wonder if there may be an industry consultant that can serve as a broker of technology and middleman. angels and VCs can also serve this purpose. again, relationship building. cant market in isolation.
Hello @Brian
I agree with you. Isolation is not a circumstance propitious to market anything. After all marketing implies an exchange between two parties. The problem becomes how to build relationships and break isolation within the given circumstances.
There should certainly be, somewhere, appropriate industry consultants, "angels" or VCs, but it is not easy to obtain their attention. Not at all.
In some third world countries academic entrepreneurship may exist, but only as another idealized concept, like democracy or justice. When action has to be taken, everything is slowed down and drowned in bureaucracy, lack of funds and all sorts of other difficulties.
For the sake of concreteness, make the assumption that the paper
https://www.researchgate.net/publication/258247920_Polyhedrons_and_Perceptrons_Are_Functionally_Equivalent
is the theoretical starting point for the NNNP.
The article proves that perceptrons (neuroscience entities) are equivalent to polyhedrons (geometric objects). And they can be easily translated into one another. Hence, a manner to obtain a perceptron that recognizes given data vectors consists in geometrically specifying a polyhedron that contains the example vectors and leaves outside the counterexample vectors. From the given data such specification can be made in an efficient manner (an algorithm exists but is not explained in the paper). And furthermore the margins (distances from the data to the decision boundaries) can be set in advance, up to the maximum theoretically possible limit. Margins are to perceptrons like ailerons to airplanes. Think of currently existing perceptrons as simple kites, not airplanes. The future is the efficiency of the perceptron specification PLUS the margin control.
Many aspects of the alluded algorithm can be the subject of a patent. But details must be kept undisclosed until confidentiality agreements and adequate financial compromises have been signed. Afterward hired experts, including patent lawyers, should prepare the description, claims and further details of the patent. Experts are expensive.
As you said, moving to the first world could be an option. But this is also an expensive step that requires financing.
If your suggestion is to be followed, and taking into account the need of undisclosed details, the problem can be reformulated as follows:
HOW TO ATTRACT THE INTEREST OF VCs TO MARKET THE DESCRIBED NNNP?
Best regards,
Daniel Crespin
angels are a type of (early stage) VC.
"Experts are expensive. "
then you have just confirmed to yourself that you need an angel or VC.
all marketing has budgets. you will need (to spend) something to get your idea going. i.e. it is unlikely going to cost you nothing. so you are inquiring about the best - cheapest, effective - way to go about.
you are also asking: what to do when your national VC would not be able to support and carry your idea. you have to ask whether your local VC market can carry your idea. some ideas are simply to big for the local VC market. then you have to find a more developed VC market.
build a network. linkedIn is cheap and can be very effective. i think you simply need the correct search keys.
also study the (your) industry. most industries have 'gatherings' trade shows, etc. that is the kind of places you wish to be. all industries have support structures too. identify experts in the industry, and connect with them, whether online or personally. in short, get closer to the industry. network(ing).
in my opinion, you can find a VC online (on linkedin) in a remote country. some VCs dont mind remote opportunities.
i would not call it that.
start conversations with local and international VCs and angels, even/ also online.
some may direct you, and give guidance, and you would learn in the process and can then adapt your strategy.
i think you are simply now coming to terms with the fact that entrepreneurship/ venturing takes a lot of effort and knocking on doors.
if you give up this easy, it is not really what you want to do. your heart is not in it.
there is not really such a thing as venturing from a basement.
Thanks for your interest and helpful ideas. A message was sent via RG.
Best regards,
Daniel
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence