Welcome to Tokeya

Tokeya is a young company located in Würzburg, Germany, offering services in the fields of

We focus on time series analysis with articfical neuronal networks and methods of the system dynamics.

As a first taste of our range of services you can play a little game against our turtle Toki.

Have a look at our website and our offer and feel free to contact us. We are happy to hear from you!

News & events

  1. Funding application to the BMBF for early detection and individual monitoring of Alzheimer's dementia using AI methods

    We are pleased to announce that Tokeya Deep Data Dive GmbH & Co. KG in cooperation with the Chair of Clinical Psychology and Psychotherapy (KliPs) at the Friedrich-Alexander-University Erlangen-Nuremberg (FAU) has submitted an application to the Federal Ministry of Education and Research (BMBF) for the early detection and individual monitoring of Alzheimer's dementia (“IASON”).

    The content of the grant application relates to the early diagnosis of Alzheimer's dementia through the combination of neuropsychological tests with advanced EEG time series analysis and the use of an intelligent emotional-empathetic digital assistant (“IEEDA”) to monitor the course of the disease through continuous speech analysis in the patient using AI methods.

  2. MoMoCa: Demonstration of an online dementia screening test
    Tokeya Deep Data Dive GmbH & Co. KG is developing an online test for research purposes to assess the cognitive ability of potential dementia patients: to the MoMoCa test
  3. Application for funding from the BMWi for adaptive load control using AI methods and line regulators

    We are pleased to announce that Tokeya Deep Data Dive GmbH & Co. KG and A. Eberle GmbH & Co. KG, in cooperation with Frauenhofer ISE, the Chair of Energy Systems and Energy Management at Kaiserslautern University of Technology and innogy SE, have submitted a joint funding application to the German Federal Ministry of Economics (BMWi) entitled “LVFACDS - LowVoltage Flexible Alternating Current Distribution Systems”.

    The content of the grant application refers to the development of an intelligent load flow control system with prediction of the state of the power grid by means of AI methods in order to ensure an intelligent adaptation of the load with increasing electromobility and the use of power flow controlling line regulators for the adaptive operation of low-voltage grids.

  4. Alzheimer dementia research using AI: analysis of EEG data

    Tokeya Deep Data Dive GmbH & Co. KG is investigating EEG data from patients with MDD (major depression disease) and AD (Alzheimer's disease) using methodsfrom AI, time series analysis and the theory of dynamic systems. The aim of these analyses is to develop methods that considerably improve the early detection of Alzheimer's dementia and at the same time provide a differentiation of age-related depressions in order to prevent mistreatment.

    In the following pictures, the first picture shows the course of the EEG for one electrode at two test persons. The second picture shows the associated “Higuchi Fractal Dimension” (HFD).The lower the HFD, the smoother the EEG raw data curve, even if it has more outliers than the reference curve. If, for example, the HFD in depression patients has a value below 1.6, this is significant for the presence of MDD. In Alzheimer's the same applies modified: the lower the activity, the smoother the curve and the lower the HFD.In this way, a complete EEG data analysis using a classification with advanced AI methods provides an early diagnosis of the presence of the disease.

  5. Application for funding in the context of Horizon 2020 for the Health Index of transformers in the electricity grid

    We are pleased to announce the application for funding in the context of the EU's Horizon 2020 programm, sub-programm SMEInst, on the subject of “Extending the service life of transformers in the electricity grid” in cooperation with A. Eberle GmbH & Co. KG and bitnova SRL (Bucharest, Romania).

    The content of the application for funding concerns the joint development of a TFIHM (Transformer Fleet Intelligent Health Manager) using methods of AI making an assessment of the current health status of transformers and a prediction of their remaining lifetime

    .
  6. Development work in the field of image recognition for autonomous driving completed
    We are pleased to announce that we have successfully completed development work for a major automotive supplier to its great satisfaction. The recognition rate for image recognition in the context of autonomous driving was significantly increased by an intelligent statistical procedure for selecting the most suitable samples for learning a CNN (Convolutional Neural Network) from a large sample set, while at the same time reducing the learning sample set to one third of the original sample set.
  7. Data privacy statement
    The current data privacy statement is available in the „Contact“ menu at „data privacy statement“. It is available only on German.
  8. Project PoPAI for A. Eberle GmbH & Co KG successfully completed

    We are pleased to announce that the project PoPAI (Power Prediction by AI) was successfully accepted by the company A. Eberle GmbH & Co. KG Nuremberg on 29th March 2018.

    The PoPAI project dealt with the prediction of important parameters in the electricity grid such as voltage and frequency. For this purpose, standard models such as ARIMA and artificial neural networks were implemented. With a practical graphical frontend (web interface) the parameters of the prediction (period of the prediction, retrospective time lag window, data resolution, …) can be set and the results can be visualized. The quality of prediction was good, best was the prediction of the neural network.

    An example prediction by the neural network (red) for 6 hours
    and for 48 hours

  9. Order from A. Eberle GmbH & Co. KG Nuremberg
    We are extremely pleased to announce our first order for 2018. We will be responsible for the company A. Eberle GmbH & Co. KG Nuremberg to implement the introduction of AI. The focus here is on devices from A. Eberle for the measurement of power quality parameters of power grids. AI is used here for time series prediction. By means of neural networks, qualitative changes in supply voltage and supply frequency are to be detected as well as faults in the networks are to be located.
  10. Talk by Dr. Thomas Fritsch at the 8th Regensburg Transformer Symposium

    At the 8th Regensburg Transformer Symposium Dr. Thomas Fritsch will give a talk entitled „On the Current State of Artificial Intelligence (AI) and its (possible) Use in the Energy Industry“. The abstract to this can be found after the links to the organizer and the conference schedule.

    Organizer: DTC Daemisch Transformer Consult

    Schedule: Draft of the conference schedule (only available in German)

    Speaker: Dr. Thomas Fritsch, CEO of Tokeya Deep Data Dive GmbH & Co. KG

    Co-Speaker: Dr. Frank Wirner, CTO of Tokeya Deep Data Dive GmbH & Co. KG

    Abstract:

    The near-perfect concurrence of 4 fundamental technological trends – very fast (also parallel computers; very high communication and storage density (smartphones, tablets etc., cloud storage & computing); very large amounts of data (Big Data - e. g. from sensors) and very effective new algorithms (Deep Neural Networks) – has since 2012 experienced a breathtaking upturn in artificial intelligence (AI). Since then, the AI has penetrated important areas of society all over the world at an ever faster pace.

    The talk will highlight the current state of development of the AI and explain why reporting on it is not a hype (this time). The working method of the main carriers of the AI, the (deep) neural networks (NN), is explained by means of practical examples from the energy technology and economy. It is made clear that this technology is a very efficient, because adaptive technical implementation of well-founded (complex) statistical learning procedures, which cannot be equated in any way with the known “expert systems” (and their weaknesses).

    The topic of predictive maintenance of important equipment of the network infrastructure (e.g. of power transformers in the power grids), which is of interest to practitioners and participants of the transformer symposium, will be treated by means of an online software tool for the intelligent condition assessment of anonymized transformer data of an industrial partner as an example for the possible efficient use of AI methods in transformer diagnostics.

    The talk concludes with an outlook on the presently foreseeable future development of the AI without succumbing to the usual and common threat scenarios or euphemisms of the potential actual dangers of the AI.

    Slides of the talk: PDF Download (only available in German; new page 11)

    In the talk addressed video:   Google’s self-driving car

  11. Visit of the GPU Technology Conference in Munich
  12. Visit of the AMA innovation forum “hardware meets software” in Dortmund