THE TRI-MOTOR IS THE VIRTUOS CIRCLE THAT DRIVES
Our Work
We are a group of scientists and developers passionate about creating artificial intelligence tools that benefit society. We are interested in projects that represent a general advance for users, companies, institutions, governments... in short, for everyone!
Q.Link.it
Qlink.it is a web implementation of Quantum Criptography algorithm (only the algorithm!). It can be used to send secure messages through any plattform.
Pharma Company
In this work, we employed data analytics to identify fraudulent physician-patient interactions. We computed probability distributions of patient variables and identified physicians whose patients deviated significantly from these distributions. Investigation of these outliers revealed valuable insights appreciated by the Pharma.
Al Dashboard for Gender violence en CongressWomen
Al Dashboard In response to the concerning prevalence of gender-based violence, we were commissioned by the National Congress to develop a Smart Board. This innovative tool tracks, classifies, and displays instances of online aggression directed towards congresswomen. By leveraging intelligent vectorization techniques on a comprehensive database, the Smart Board analyzes the language used to identify and categorize these gender-based attacks. This data visualization serves to raise public awareness regarding the significant issue of online harassment faced by female representatives.
IntelSocial
We serve as the primary AI service provider for IntelSocial (accessible via @intelsocialbot on Telegram). Our system ingests data from over 1,000 Argentinian newspapers and relevant social media platforms on a continuous basis. By applying multi-level, in-depth AI analysis, we develop innovative tools and indicators to enhance public discourse. One key capability is the detection of coordinated news dissemination (mono-editorialization) across media with similar editorial stances. Additionally, we generate heat maps visualizing geographical patterns of news coverage across Argentina. These regional patterns can reveal potential biases or trends related to specific events, such as surges in news related to insecurity or traffic accidents. Ultimately, our tools empower policymakers with data-driven insights to optimize public policies.
Al @ Cryptos
We explored the short-term cryptocurrency price prediction within the high-frequency trading domain using LSTM networks. We established error and uncertainty metrics to ensure the reliability of the predictions. Our findings revealed that the model could accurately predict price movements exceeding 3 standard deviations only if they were below the Binance fee. For larger variations, the accuracy significantly decreased, which we termed the "Market Uncertainty Principle." The most valuable outcome for the consortium was the confirmation of the limitations of the approach. This allowed them to make an informed decision to exit the cryptocurrency market at an opportune time.
Al-Friendly.com
Al.Friendly.com is an open educative plattform in which users can easily understand and train a Neural network for the first time.
Al Assessment
Muchas organizaciones reconocen el potencial del Aprendizaje Automático (AA) pero tienen dificultades para dar el primer paso. Nuestro servicio funciona como el punto de partida ideal. Trabajamos bajo Acuerdos de Confidencialidad (NDA) para obtener una comprensión profunda del panorama de datos de su organización. Esto incluye métodos de adquisición de datos, flujo de datos, utilización actual y potencial sin explotar. En pocas semanas, entregamos un informe completo que describe no solo una variedad de aplicaciones de IA prometedoras sino también recomendaciones para optimizar la adquisición y el flujo de datos. Proponemos algoritmos de AA específicos adaptados a sus necesidades, junto con estimaciones expertas de su potencial de éxito basadas en nuestra amplia experiencia y conocimiento. Además, determinamos el impacto potencial de cada implementación propuesta y estimamos los requisitos de recursos (hardware, software, experiencia en IA y horas de desarrollo). Este informe personalizado le brinda a su organización una hoja de ruta estratégica para aprovechar el AA de manera efectiva.
Al for Executive Leaders
We designed a 9-lecture course for executive leaders (VP and above) with the objective of introducing them to the main tools within the disciplines of AI and Machine Learning to lead teams and make decisions. The goal of each customized lecture is to understand AI without the need to see or write a line of code, by understanding the main algorithms, tricks, purposes, scope, possibilities of success and failure, and intuition about data. This training is highly effective and has been taken by high-level Argentine executives and large companies.
Intelligent Socio-Economic Index
Socio-Economic map for Buenos Aires. We observed a non-random structure in responses to a comprehensive census survey in Buenos Aires, where all 79 questions per household exhibited a strong socioeconomic bias. Recognizing this underlying latent variable, likely representing socioeconomic status, we employed matrix factorization techniques to infer its continuous distribution across the city. This unsupervised approach yields the first continuous heatmap of socioeconomic levels for Argentinian cities, offering a crucial advantage in downstream analyses compared to traditional discrete or questionnaire-based methods.
COVID Phone-calls Early Alarm
We explored the relationship between patterns in 911 call data and confirmed COVID-19 cases. By analyzing the volume and origin of these calls, we developed a software tool capable of estimating the number of COVID-19 cases. This work is further substantiated by a scientific article published in the prestigious The Royal Society.
Voice-to-Text
We have developed a proprietary, server-based Spanish language voice-to-text (STT) system. This system is currently utilized within IntelSocial (@intelsocialbot, testable via "/radios" command). Here, it serves a multifaceted role in monitoring radio and TV broadcasts, including tasks like verifying advertising compliance, identifying non-traditional marketing strategies, and analyzing news coverage of specific entities (people, companies, etc.).
Basic research & academy
Our commitment to foundational research and AI curriculum development fuels our ability to stay at the forefront of the field. This continuous learning process allows us to integrate new techniques into our application and tool creation. We further augment our expertise by organizing workshops on specific areas of interest for knowledge expansion. Rooted in academia, our team boasts a strong university background.Explore some of our research papers utilizing AI for fundamental research purposes:
Link 1Link 2Link 3Link 4Link 5Link 6
