Topic: [ANN] Adding AI to every website (Read 1716 times)

What is the key idea of the project?

i cant say that their product is very best in market. But i can trust to that company, because i started website creating with them, and they still working and growing

Aesthetics is a gradient boosting model which has learned information about 10,000 websites and about 100,000 their ratings. It accepts features from two sources: the image of the website and its source code. This hybrid feature extraction provides 125 attributes.

In addition, there's an "optical" high-dimensional CNN feature extraction that is at the experimental stage.

The process of model calibration had gone through several iterations before the quality of the evaluation was achieved. An average error is 0.5 points per site in the ten-point system. By the way, our assessors made mistakes of about 1.5 points per site, which is worse than our model.

Sorry, I don't buy story about your expertise. Mostly because:
1. You openly lied at least once about yourself (that's enough to assume anything else you say is a lie);
2. 8 month is like nothing in data science;
3. link your provided for website evaluation "tool" does not show anything except what? A progress bar that states "Aesthetics of your website is arbitrary %" - that's not data science - that what kids do after school. 

• $200 price of one token
• 100 000 UKT tokens will be issued in total
• 55 000 UKT tokens will be sold during ICO and pre-ICO

For more information read our WP: https://ico.ukit.com/download/ukit-ai-whitepaper_en.pdf (page 30)

1. It is the appearance that the tool checks. The code validity and other aspects are not evaluated but if the website collapsed because of errors in the code, the rating will be low, which is logical.

2. Yes, sure. Now we really need additional ratings and seeing if our ratings are correct in your opinion. Your rating will serve to improve the data set and finish the training of AI. Although it won't be taken into account in real time, it'll be kept by the system for further use.

3. Now showing personal and overall search histories seems the most interesting.
But this is mostly for the demonstration and for the opportunity to quickly assess the correctness of work. Remember, this scoring tool is meant for a different purpose in the first place.

https://aesthetics-demo.ukit.space/ is a pretty interesting tool. Here are a few questions about the usage of AI itself in the project:

1. Is the rating based on the appearance and/or code of the website?
2. Does the user rating (if I provide my own rating) influence the training of AI?
3. And the question about the history of recently evaluated websites. Will it always be accessible to everyone? Or will there be a private/personal option?

In addition to Roman’s general background, such as theoretical knowledge and being a Ph.D in Physics and Mathematics as well as an Associate Professor at the Institute of Mathematics, Mechanics and Computer Sciences, Roman has been working on machine learning tasks for 8 months as part of the uKit AI project.

And again, like I said, our company has ready and operating products under the belt — two website builders and the basic foundation for the future uKit AI system. All this proves the background of our big team consisting of over 100 people, rather than that of one specific specialist.


If you have a look at our Whitepaper, we are planning to spend a significant amount of funds on R&D. This means that in the first place we’ll be strengthening our team, and, of course, the machine learning part will be the first to be enhanced.

Additionally, not only do we have the team building experience, but also best practices in creating actual, working products from scratch, which lets us assess our expertise in one way or another.
Plus, you might be interested to see the screenshots of our uKit.ai 1.0 “lab”

http://screenshot.ukit.com/src/viger/1.png
http://screenshot.ukit.com/src/viger/2.png


They show that even now websites are successfully taken apart into components and how this process looks like though computer vision.

This is the system that already exists — uKit AI 1.0. It hasn’t been released yet, but it has been already trained and unveiled within the team. The system can mark meaningful blocks on the website, detect their theme, identify what kind of the block they belong to, and structure these data to further transform the website into a new one in our uKit website builder.

But, of course, it’s better to see this in dynamics, and we’ll definitely make screencasts so that it would be more clear.

And since hearing a hundred times is not as good as seeing once, and taking into account that you, Colorblind, have clearly paid a lot of attention to us. Btw, a special thanks to you for this. Your sharp questions clearly help us look at ourselves and the provided information with a critical eye. We just wish you had more time to look into all the answers and references that we share. But I understand that it’s not really easy to do that, and it’s not your job to do thorough researches.

So, about seeing once, judging by your posts we have a feeling that at the end of November you are potentially planning to visit Moscow for the blockchain conference. We’ll be happy to become your host and specifically invite you to our office, show you our developments, and if you are interested also have a chat with our technical specialists including Roman. This way, you’ll be able to form a truly objective opinion. I think it could be interesting for everyone, and you have already delved deep enough to take another step.

Sorry for not getting back to you sooner. I should have replied to you earlier, but our conversation seemed to fade down and I still didn’t hear what could serve as a real illustration for you.

Of course, the main background is neural networks and machine learning algorithms. As I have told you, my previous hands-on experience is protected by NDA. If you want to have a look at it, I can share it via a personal message. The company is CVisionLab («CиBижинЛaб»), you may call them for a confirmation.

This agreement is in the Russian language as it is concluded with the Russian company. To check it, it’ll be enough to contact them at the provided corporate details. If someone is really interested, bitcointalk has a significant Russian-speaking community, you may request assistance from some of the authority figures there.

For 8 months of work, a number of developments has been done at uKit. I’m afraid, there are very few areas where fast results are possible with machine learning. Speaking about something you can see in action and “touch”, here https://aesthetics-demo.ukit.space/

At the moment, https://aesthetics-demo.ukit.space/ looks more like a toy that makes it possible to evaluate your or another person’s website. But we meant it this way so that people had something to ‘touch’. Also, through gathering opinions of a wide variety of people and offering them to correct our ratings we are able to collect a bigger data set.
Basically, this is the scoring system for uKit AI and is a subsequently important, yet supplementary part of the product.

This system is built using machine learning methods. It should reflect an opinion of an average user about the website’s appearance. It’s neither about a professional designer’s opinion, nor about the website’s usability, nor about its functionality. Let me emphasize this point — it’s only about the APPEARANCE and only an AVERAGE user.
In this project, we are using one of the variants of the gradient boosting method (catboost library). To date, this approach is one of the most effective in solving problems of machine learning.

Besides, in the near future (this month), we are planning to introduce an additional evaluation by the neural network, which should improve the overall quality.

To train the model, we have gathered a sample of 12K websites. Despite the fact that at the moment the system gives reasonably good results (it is possible to play around with it on your own), it is still the first alpha version and has a number of issues:

• it can’t evaluate minimalistic websites such as google.com (this is a solvable issue, but we’d like to achieve this through machine learning rather than using “white lists”);
• it can’t work correctly with hieroglyphic websites (Chinese, Japanese…), since there were none of this kind in the existing training base;
• it has an insufficient ratings bank. We need more people and different opinions to improve the quality. It doesn’t make the system unrepresentative at the moment, but the data set needs significant enhancements, which will certainly influence the quality and robustness of the output results.

it seems something managed to stop him after all...

And I read all your questions and doubts. And if you look from the outside, they are fair, and you are right, Roman gave just a general words. If I did not know him in person and did't work with him - that would not satisfy me either. And he, as a person who was used to the fact that his authority is known (including within his teaching activities) could not respond normally.
I think tomorrow nothing will stop him from continuing the dialogue in a constructive way and give a real proofs of his expertise.

It seems that you’ve never been interested in researching website builders
First off, there are a few things to consider:
We have 2 main independent products in the website building niche: uCoz and uKit. Both are already popular on mass market, though uKit is fresher and more modern. Basically, for us it’s a logical spin-off of the good old uCoz, but having a different audience and meeting present-day needs. They have different goals, target audiences and user bases.

You might not have heard of them as our main territory is Eastern Europe. But it doesn’t mean that there’s no information about us in the English-speaking segment. You can easily google us and find reviews about our website builders, guides, testimonials and so on, they have been around for a long time.

There are also English-speaking themed top lists of website builders that name us one of the best products on the market, for example:
https://superbwebsitebuilders.com/
https://www.pcmag.com/article2/0,2817,2484510,00.asp

There were big milestones in the history of our company:
- A deal with Yandex about their free hosting narod.ru
- An investment round (a while ago) from DST (also famous for their investments into Facebook, Groupon, Zynga, Mail.Ru). And, of course, throughout its history, our company has gone through due diligences way more serious than that you are conducting now based on forum answers and your personal judgement.

You can read about it here, for example:
http://www.ewdn.com/2013/02/18/yandex-to-transfer-website-hosting-service-to-ucoz/
I’m all for objectiveness and facts. You take everything with a pinch of salt, so let’s clarify things — it’ll be right and fair.


https://www.comscore.com/Insights/Press-Releases/2011/11/comScore-Releases-Overview-of-European-Internet-Usage-in-September-2011
Here you can see that our service has been around for a while and was featured by big and reputable analytic companies.

As far as the examples of users’ websites go, I’m afraid, there’s no way to show all our websites following one specific footprint with one request. We use different brand names for our products (in one product) and image logos in site footers as anchor links. The system’s copyrights are removed when users buy premium packages and in case of uKit even without purchasing a plan.


Surely, on our side we can export the list of all websites that we serve, but, as you can understand, this will be a dream of search spammers come true, a gift, not to mention the ethical side of the matter in connection with our clients, and our trade secret.

But certainly there are third-party services that make it possible to measure the amount, find and view a part of it:

https://w3techs.com/technologies/details/cm-ucoz/all/all
https://w3techs.com/technologies/history_overview/content_management/all


https://statonline.ru/domains?tld=ru&dns=ns2.ucoz.net
https://statonline.ru/domains?tld=ru&dns=ns1.ukit.com
(We have to understand that this is only one domain extension, one connection method. We offer to connect a domain via Name Servers or IP pointing, and only paying users can have a custom second-level domain.)



Honestly, I can’t agree with you. Since I spent a while building our Neural Network team, small so far though, I had an opportunity to research and communicate with all kinds of specialists. I am sure that our guys have a strong basis, and, more than that, have experience in production, while in this sphere you typically come across people with only theoretical background without its practical application.

But certainly, since ICO is about fundraising, and the main tasks lie exactly in the ML field, we will need to grow and empower our team, and a significant part of funding will be spent on this.


I’ll be happy to answer all the questions you might have. And I will appreciate if you read all the information. I just have a feeling that you tend to grab pieces from here and there.

All there is a linkedin profile of someone who allegedly worked some 7 month in company that does image recognition. No real projects link (ah, ofc, they are under NDA), no articles, no codebase only a lot of words and of course a lie (why would he lie? This really bothers me). Also he mentions some girl who won scholarship isn't listed in your team (perhaps you not listing her because she will be soon off to actually study?). He also mentioned he was invited to work in Russian search engine Yandex (three times!) Was it also a lie?

If you visit our site - ico.ukit.com (it was here from the very beginning of this topic, we didn't hide it ^)) you are able to find live links to the members Ln profiles in Our Team section).

I posted simple request - provide background of your data scientists. Link their social profiles, articles and graduation papers. Instead one of them came here and openly lied to my face (twice actually). Also you have posted two videos with little to no information about what they are (one sounded like neural network trying to speak and failing) the other one is just something. But I really more concerned about lies of your team member. If he lied about one thing, could it mean he lied about other things as well? Why would anyone trust you after this?

In the first place, I'd like to note that that  Cosmics Project is not a part of our current uKit AI project — it's the background since you doubted the qualifications and level of experience in this field. The project is long-term and if you're interested in discussing it further and seeing what's under the hood — welcome, it can be set up. But I guess it'll be better to do this in a separate topic.

Well, nobody claims that the invented AI describes the specific research.

As far as its validity goes, using tensorflow it's impossible to create a physical simulation or a visualization that is shown in the video. It seems you don't fully understand what you are talking about.

If it's a question of trust, I can show you dozens of runtimes of completely different simulations and their source codes, including the process of launching these simulations and demonstrating the source code. It would be quite difficult to fake this.

And if you are curious to see the proofs, let me know what you'd like to see to clear up your doubts. I have no problems showing them to you.

Of course I will use word SCAM.
Your team member openly lies about his background (see my previous post).
Do you think it is good for your project?

You seem not to understand the concept of our relations. I don't have to prove my expertise to you or explain anything about myself. I can be newbie or hero member - does not matter. It is You who should prove your expertise to the community because it is You who ask for money.

And you are failing.
 

Yes, I know than a word "scam" could appear in each thread In our case I would understand if you said that you don't believe in our idea, or you doubt that we will really be able to finish the project. But blame us as a scam…
We are based on our existing projects (ukit.com), users base and reputation. (That’s really easy to check on the Internet)


Now I started to think that you are not really interested in the dialog or our project whatsoever and just trying to add more posts to your Newbie status.

Oh Russians are always so proud

first of all you, sir, are LIER because



is a lie. It clearly states that ALL tuition being done in evening, and will require no more than 21 hours/week (basically 3 hours/day) and can be done from anywhere.
https://yandexdataschool.ru/admission/faq .Google translate gives pretty straightforward translation:


This is the second time I'm catching you on your lies. That's bad. Post me a link where it "DEMANDS" you being out of full time job and I'll apologize.



You see, Roman, you may like or not like my style but you came here to ask MONEY from people. And investors will not ever care about your feelings. When you ask for money you have to be prepared to hear a lot of very uncomfortable questions and be prepared to answer (for example - and this is what you should have been done few hours ago - provide actual links on your social network profiles). At this point you have provided nothing about your background besides general words. It's not me who should ask you for your profile - it is you who should showcase it, yet you still only posted your name, not links for your profiles.

  

This forum is known for a lot of fake projects and scams.
This is why people here are little more than paranoid about their investments.
And you are doing pretty bad right now.

Well, I said that I can't afford it because I have to work. School of Data Analysis demands candidates should be out of any full time job. I know the person who was not allowed to pass the exams for this reason. And I mean I can't study for 2 years without earning money. If it is a reason to blame me, then I think you have to blame 90% of adults (may be less may be more).

I'm not hiding my identity in this topic. I shown up there in the team section. My name is Roman Steinberg. So, you can look at my LinkedIn, github, WebOfScience profile or whatever you want.

We both are newbies on this forum, but I don't like your style of dialogue. I tried to answer in general words. If I misunderstood your level of knowledge in the area then may be you didn't shown it first. Now you blame me that I'm hiding something. No, I just tried to answer generally. Good luck in your investments!