Making deep learning learn faster

We offer technology called Transfer Learning on Steroids (TLoS). With this technology deep networks can learn much more efficiently than by using traditional methods.

If your deep learning model does not train as well as you would like it, or deep learning even cannot be applied because you have too small training data sets, we may have a solution for you: we offer a technology for training deep learning models that is considerably more efficient than anything else out there.

How does TLoS work?

Our training platform lays a “knowledge blanket” that imposes inductive biases over a deep learning network. As a result, learning becomes orders of magnitude more efficient than by other methods.
UPCOMING Agent Store
How to use Mental AI Platform?

Pick the matching MENTAL agent from the agent store that best addresses your problem. If you haven’t found a matching one, you may want to build your own agent and make some money by doing that (expected Q4 2020).

  • Upload your model

    We apply our technology on top of your existing models source code, while supporting all the major ML frameworks.

  • Upload your data

    We support all major industry standards for dataset formats.

  • Wait a few hours

    Wait while we apply our MENTAL enterprise-/military-grade magic and train your model.

  • Use it and integrate

    We provide trained mental agent for download with best practices on how to integrate it further.

The Pricing Model
for Mental Agents
Operations: 100K
Mental API

  • Monthly
  • Response Time: Slow
  • Integration and Support: No

We charge for a specific activity’s per: training epoch or inference call, as API operation.

Operations: 1 Mio.

  • Monthly or yearly
  • Response Time: Medium
  • Integration and support: No

We charge a fixed amount per month/year per agent. The no usage limit during this period.

Operations: UNLIMITED
Instance on Go Mental
AI Platform

  • Monthly or yearly
  • Response time: Fast
  • Integration and Support: Yes

The client uses or creates agents on their own instance based on Docker containers.

The prices will be set in Q3 2019.

Enable the world to train deep learning networks with small data sets.

Some researchers believe AI could catch up to human intellect if it starts off learning like humans do—as babbling, aimless toddlers.

Kindergarten for Robots Is a Lot Like Kindergarten for Kids

“The Future Technology of Artificial Intelligence” for inTAIL 2018 (video interview) – Intelligent Retail at ELSE Corp.

inTAIL 2018 –The Future of Artificial Intelligence

Danko Nikolić hat einen “Kindergarten für künstliche Intelligenz” entwickelt, der Computer so schlau machen soll wie Menschen.

Maschinen das Träumen lehren