Explore the non-commercial data and apps:
EXTRACTING EMOTIONS FROM VOICE SIGNALS
In this example, the avatar mimics the emotions captured by the user`s voice. It uses a combination of SVM algorithm and Decision Trees with Context-Oriented approach.
You will note that the avatar tends to select one of the expected emotions for that context. The avatar doesn't need a strong confidence to select one of them because they are "expected" for that context, in order of relevance.
It is still it is open to other emotions when it identifies a strong correlation between the voice signal and other emotions. This approach increases the accuracy, in especial for borderline voice samples.
CROWD SIMULATION IN EVACUATION SCENARIOS
In this example you see Crowd Simulation using Network Oriented model to simulate the mind model, social and cultural aspects of each agent and the influence of these aspects on evacuation dynamics.
MIND MODEL OF TRAUMATISED PEOPLE
What can we expect about the recovery of traumatised patients?
What is the impact of group treatments?
Normally, patients take years to overcome traumas and data of long periods are scarce. These factors make hard answer the above questions. Fortunately, in last years, many advances in neuroscience allows us to develop mind models about the emotion regulation mechanisms of brain.
In papers 2016-1 and its extension 2017-7 your can find literature of those brain mechanisms and its translation to network-oriented computational model, focused on traumatised patients. The model can also emulate “allostasis” effect introduced by McEwen, in his work 'Physiology and Neurobiology of Stress and Adaptation: Central Role of the Brain'.
Press the button to access the source code in Matlab and Database used to validate the model.
DATA MINING TO DETECT FRAUDS
Frauds always left finger prints and datamining algorithms are a powerful tool to detect them.
Here you find the source code in python of a system that uses DBSCAN algorithm to detect frauds in elections. The system has a web interface and analyses the voting in micro-regions, cities and regions, suggesting suspecting fraud patterns. The description of the system can be found in Paper 2015-1.The source code is available by clicking on bellow button. For more information, please contact me or Yuri Poloni, the main author of the system.
PROTOTYPE - SMART SYSTEM TO DETECT DEPRESSION IN ELDERLY
A brief overview of smart home system that identifies initial stages of depression and make interventions to reduce the risks of depression in lonely elderly.
You also will find an incipient version of the code developed to Arduino board and Telegram.
Authors: Daan Beverdam, Daniel Formolo, Gossa Lô, Matteo Pullano
PROTOTYPE - SMART ENVIRONMENT HOLOCONSCIOUS
HoloConscious is an intelligent system that monitors unwanted events when dwellers are out of home or in situations of elderly living alone. The system avoids dangerous situations for dwellers and their neighborhood, saves energy and generate a huge database of habits without cross the border of privacy. To promote these benefits, the system sends notifications to dwellers and takes decisions/actions itself to prevent accidents.