Juan Pablo Zuluaga

Juan Pablo Zuluaga
Idiap Research Institute | IDIAP · Computer Science

PhD Student

About

55
Publications
30,385
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
158
Citations
Introduction
Currently working at IDIAP as a jointly PhD student with the École polytechnique fédérale de Lausanne. Working in the European Union Horizon 2020 ATCO2 project, dedicated to the development of an Automatic Air Traffic Speech Recognition system. ATCO2 aims at developing a unique platform allowing to collect, organize and pre-process air-traffic control (voice communication) data from air space. Also interested in biomedical imaging and almost everything related to Machine Learning and AI.
Additional affiliations
January 2019 - present
Institut FEMTO-ST
Position
  • Research Assistant
November 2015 - November 2016
Universidad Autónoma del Caribe
Position
  • Research Assistant
July 2013 - present
Universidad Autónoma del Caribe
Position
  • Researcher
Education
January 2020 - January 2024
École Polytechnique Fédérale de Lausanne
Field of study
  • Automatic Speech Recognition and Natural Language Processing
February 2019 - September 2019
École Nationale Supérieure de Mécanique et des Microtechniques
Field of study
  • Nanotechnology, control systems
August 2018 - February 2019
Ivanovo State Power University
Field of study
  • Advance control systems

Publications

Publications (55)
Article
Breast cancer is a disease that threat many women's life, thus, the early and accurate detection play a key role in reducing the mortality rate. Mammography stands as the reference technique for breast cancer screening; nevertheless, many countries still lack access to mammograms due to economic, social and cultural issues. Last advances in computa...
Preprint
Full-text available
A recent study from GLOBOCAN disclosed that during 2018 two million women worldwide had been diagnosed from breast cancer. This study presents a computer-aided diagnosis system based on convolutional neural networks as an alternative diagnosis methodology for breast cancer diagnosis with thermal images. Experimental results showed that lower false-...
Preprint
Full-text available
Air traffic management and specifically air-traffic control (ATC) rely mostly on voice communications between Air Traffic Controllers (ATCos) and pilots. In most cases, these voice communications follow a well-defined grammar that could be leveraged in Automatic Speech Recognition (ASR) technologies. The callsign used to address an airplane is an e...
Preprint
Full-text available
Recent work on self-supervised pre-training focus on leveraging large-scale unlabeled speech data to build robust end-to-end (E2E) acoustic models (AM) that can be later fine-tuned on downstream tasks e.g., automatic speech recognition (ASR). Yet, few works investigated the impact on performance when the data substantially differs between the pre-t...
Preprint
Full-text available
Automatic speech recognition (ASR) allows transcribing the communications between air traffic controllers (ATCOs) and aircraft pilots. The transcriptions are used later to extract ATC named entities e.g., aircraft callsigns, command types, or values. One common challenge is Speech Activity Detection (SAD) and diarization system. If one of them fail...
Preprint
Full-text available
In this paper, we describe our shared task submissions for Subtask 2 in CASE-2022, Event Causality Identification with Casual News Corpus. The challenge focused on the automatic detection of all cause-effect-signal spans present in the sentence from news-media. We detect cause-effect-signal spans in a sentence using T5 -- a pre-trained autoregressi...
Preprint
Full-text available
In this paper, we describe our participation in the subtask 1 of CASE-2022, Event Causality Identification with Casual News Corpus. We address the Causal Relation Identification (CRI) task by exploiting a set of simple yet complementary techniques for fine-tuning language models (LMs) on a small number of annotated examples (i.e., a few-shot config...
Preprint
The data imbalance between air traffic controller (ATCO) and pilot leads to pooling both data to train an automatic speech recognition (ASR) for air traffic communication. This often provides better performance for the ATCO than the pilot. In our previous work, we showed that in noisy conditions training ATCO and pilot acoustic models separately ca...
Preprint
Full-text available
Automatic Speech Recognition (ASR), as the assistance of speech communication between pilots and air-traffic controllers, can significantly reduce the complexity of the task and increase the reliability of transmitted information. ASR application can lead to a lower number of incidents caused by misunderstanding and improve air traffic management (...
Article
Full-text available
This document describes our pipeline for automatic processing of ATCO pilot audio communication we developed as part of the ATCO2 project. So far, we collected two thousand hours of audio recordings that we either preprocessed for the transcribers or used for semi-supervised training. Both methods of using the collected data can further improve our...
Article
Full-text available
Phonological-based features (articulatory features, AFs) describe the movements of the vocal organ which are shared across languages. This paper investigates a domain-adversarial neural network (DANN) to extract reliable AFs, and different multi-stream techniques are used for cross-lingual speech recognition. First, a novel universal phonological a...
Conference Paper
Full-text available
Contextual adaptation of ASR can be very beneficial for multi-accent and often noisy Air-Traffic Control (ATC) speech. Our focus is call-sign recognition, which can be used to track conversations of ATC operators with individual airplanes. We developed a two-stage boosting strategy, consisting of HCLG boosting and Lattice boosting. Both are impleme...
Conference Paper
Full-text available
Air traffic management and specifically air-traffic control (ATC) rely mostly on voice communications between Air Traffic Controllers (ATCos) and pilots. In most cases, these voice communications follow a well-defined grammar that could be leveraged in Automatic Speech Recognition (ASR) technologies. The callsign used to address an airplane is an e...
Preprint
Full-text available
Automatic Speech Recognition (ASR) can be used as the assistance of speech communication between pilots and air-traffic controllers. Its application can significantly reduce the complexity of the task and increase the reliability of transmitted information. Evidently, high accuracy predictions are needed to minimize the risk of errors. Especially,...
Preprint
Full-text available
Assistant Based Speech Recognition (ABSR) for air traffic control is generally trained by pooling both Air Traffic Controller (ATCO) and pilot data. In practice, this is motivated by the fact that the proportion of pilot data is lesser compared to ATCO while their standard language of communication is similar. However, due to data imbalance of ATCO...
Preprint
Full-text available
Assistant Based Speech Recognition (ABSR) for air traffic control is generally trained by pooling both Air Traffic Controller (ATCO) and pilot data. In practice, this is motivated by the fact that the proportion of pilot data is lesser compared to ATCO while their standard language of communication is similar. However, due to data imbalance of ATCO...
Article
Full-text available
Voice communication is the main channel to exchange information between pilots and Air-Traffic Controllers (ATCos). Recently, several projects have explored the employment of speech recognition technology to automatically extract spoken key information such as call signs, commands, and values, which can be used to reduce ATCos’ workload and increas...
Conference Paper
Full-text available
Advances in Automatic Speech Recognition (ASR) over the last decade opened new areas of speech-based automation such as in Air-Traffic Control (ATC) environments. Currently, voice communication and data links communications are the only way of contact between pilots and Air-Traffic Controllers (ATCo), where the former is the most widely used and th...
Preprint
Full-text available
We present a simple wrapper that is useful to train acoustic models in PyTorch using Kaldi's LF-MMI training framework. The wrapper, called pkwrap (short form of PyTorch kaldi wrapper), enables the user to utilize the flexibility provided by PyTorch in designing model architectures. It exposes the LF-MMI cost function as an autograd function. Other...
Article
A recent study from GLOBOCAN disclosed that during 2018 two million women worldwide had been diagnosed with breast cancer. Currently, mammography, magnetic resonance imaging, ultrasound, and biopsies are the main screening techniques, which require either, expensive devices or personal qualified; but some countries still lack access due to economic...
Patent
Full-text available
Muchos deportistas, principiantes y de alto rendimiento a la hora de realizar sus prácticas, carecen de equipos adecuados para entrenamientos óptimos. Debido a esto se desarrolló una maquina electromecánica que reaccionara aleatoriamente a los golpes que se logren acertar en el dispositivo y con ello permita calcular la eficiencia del deportista y...
Technical Report
Full-text available
We present a simple wrapper that is useful to train acoustic models in PyTorch using Kaldi's LF-MMI training framework. The wrapper, called pkwrap (short form of PyTorch kaldi wrapper), enables the user to utilize the flexibility provided by PyTorch in designing model architectures. It exposes the LF-MMI cost function as an autograd function. Other...
Preprint
The ATCO2 Consortium is working on the use of speech technologies (and in particular speech recognition) to transcribe pilot-air-controller conversations. It is investigating the legal and ethical compliance of ATC data collection, sharing and analysing. Air traffic control data collections are based on the setup of a radio-receiver that capture th...
Preprint
Full-text available
Voice communication is the main channel to exchange information between pilots and air-traffic controllers (ATCO). Recently, several projects have explored the employment of speech recognition technology to automatically extract spoken key information such as callsigns, commands, and values, which can be used to reduce ATCOs workload and increase p...
Preprint
Full-text available
Advances in Automatic Speech Recognition (ASR) over the last decade opened new areas of speech-based automation such as in Air-Traffic Control (ATC) environment. Currently, voice communication and data links communications are the only way of contact between pilots and Air-Traffic Controllers (ATCo), where the former is the most widely used and the...
Article
Freshwater is a critical component for social and economic sustainability, but also the water scarcity is a foremost threat nowadays. During the last century, many governments have been implementing public policies for reducing the water consumption, optimize the potabilization processes, and promote cleaner techniques for wastewater treatment such...
Article
Full-text available
The traditional detection methods have the disadvantages of radiation exposure, high cost, and shortage of medical resources, which restrict the popularity of early screening for breast cancer. An inexpensive, accessible, and friendly way to detect is urgently needed. Infrared thermography, an emerging means to breast cancer detection, is extremely...
Poster
Full-text available
Breast cancer is one of the most threatening diseases in women's life; thus, the early and accurate diagnosis plays a key role in reducing the risk of death in a patient's life. Mammography stands as the reference technique for breast cancer screening; nevertheless, many countries still lack access to mammograms due to economic, social, and cultura...
Patent
Full-text available
This paper presents the development of an accelerometer based system that captures and analyzes Matlab cardiac signals using Seismocardiography (SCG). SCG is an effective non-invasive method of signaling detection in the chest. The project was divided into three phases; the first one was the selection of an accelerometer, data acquisition system (D...
Preprint
Full-text available
Breast cancer is one of the most threatening disease in women’s life, thus, the early and accurate diagnosis play a key role in reducing the risk of death in patient’s life. Mammography stands as the reference technique for breast cancer screening, nevertheless many countries still lack access to mammograms due to economic, social and cultural issu...
Thesis
Full-text available
Breast cancer is one of the most threatening diseases in women’s life; thus, the early and accurate diagnosis plays a key role in reducing the risk of death in a patient’s life. Mammography stands as the reference technique for breast cancer screening; nevertheless, many countries still lack access to mammograms due to economic, social, and cultura...
Experiment Findings
Full-text available
Spearman Correlation between 9 features. Database: [1] J. Jossinet (1996) Variability of impedivity in normal and pathological breast tissue. Med. & Biol. Eng. & Comput, 34: 346-350. [2] JE Silva, JP Marques de Sá, J Jossinet (2000) Classification of Breast Tissue by Electrical Impedance Spectroscopy. Med & Bio Eng & Computing, 38:26-30.
Experiment Findings
Full-text available
Correlation test in Python, also seven different machine learning techniques were used to validate. Database from: https://www.researchgate.net/publication/322262037_Using_Resistin_glucose_age_and_BMI_to_predict_the_presence_of_breast_cancer
Preprint
Full-text available
Fresh water is a critical component for social and economic sustainability, but also the water scarcity is a foremost threat nowadays. During the last century, many governments have been implementing public policies for reduce the water consumption, optimize the potabilization processes, and promote cleaner techniques for wastewater treatment like...
Preprint
Full-text available
The control of mechatronic systems like position, velocity or acceleration is highly researched and helpful to understand the different kind of controllers, this article is focused in testing different types of controllers to satisfy a variety characteristics, like setting time, maximum peak, final value, rising time, etc. The first controller pres...
Preprint
Full-text available
Identification of a unknown system have been highly researched in the academical community in the last decades, such using different types of algorithm, for example in this paper is commented and also explained the advantages and disadvantages of regular scanning and monte-carlo algorithm versus genetic algorithm for identification of an unknown sy...
Conference Paper
Full-text available
This study focuses on the student's perception of Project-based learning, where this contributes to the grades of several subjects of the Erasmus Mun-dus Master in Mechatronic Engineering of the University of Oviedo. Likewise it was important to be able to analyze the improvement of teamwork skills about the groups proposed to solve various problem...
Preprint
Full-text available
The mechanisms synthesis problem is an activity that has became outstanding in the recent years, because it allows to obtain more efficient mechanisms alongside restrictions provided by the user, on the other hand it enables to minimize the system final cost and the total energy spent at the time of desired performing movement. Nowadays, synthesis...
Preprint
Full-text available
The control position of a DC motor, knowing the specific parameters of the motor, that controls a one degree freedom mechanism is discussed in the present article. The mathematical model of the motor, can be obtained from the specific parameters provided by the producer; The PI control algorithm is adopted in 3 different parameters, Proportional (P...
Preprint
Full-text available
The speed control of a DC motor without knowing the specific parameters of the motor, is discussed in the present article. The mathematical model of the motor, and the controller itself, can be obtained from practical measurement, system identification. Using a Oscilloscope and applying a Step voltage in the motor, can be found characteristics, lik...
Preprint
Full-text available
The speed control implemented on a PIC of a DC motor without knowing the specific parameters of the motor, is discussed in the present article. The mathematical model of the motor, and the controller itself, can be obtained from practical measurement, such as system identification. 1. Using a Oscilloscope and applying a Step voltage in the motor, c...
Presentation
Full-text available
La cardiografía de impedancia es un método por el cual se puede evaluar el desempeño del corazón y el funcionamiento tanto preventivo como de seguimiento del sistema circulatorio. El presente proyecto pretende deseñar un bioinstrumento para la realización de la cardiografía de impedancia para posteriormente caracterizar los puntos claves de su morf...
Presentation
Full-text available
Aproximadamente 415 millones de personas fueron diagnosticadas con diabetes en el año 2015, donde para la mayoría de personas es casi imposible financiar alternativas para mejorar su calidad de vida. Con el desarrollo de este dispositivo se quiere lograr que todas las personas diabéticas puedan tenerlo a su alcance, con el fin de que se les facilit...
Article
Full-text available
This paper presents the development of a accelerometer based system to capture and analyze in Matlab cardiac signals using Seismocardiography (SCG). SCG is an effective non-invasive method for the detention of cardiac signal in the chest. The project was divided into three phases; the first one was the selection of an accelerometer, data acquisitio...
Article
Full-text available
El presente trabajo muestra el desarrollo de un sistema basado en acelerometria, para la captacion y analisis en Matlab, de senales precordiales usando sismocardiografia (SCG). La SCG es un metodo eficaz para captar las senales en el area precordial de forma no invasiva. El proyecto se dividio en tres fases: la primera tuvo de objeto la seleccion d...
Conference Paper
Full-text available
El presente trabajo muestra los avances en el diseño y construcción de un sistema basado en acelerometría, para la captación y análisis en Matlab, de señales precordiales usando sismocardiografía. Para la medición fue necesario establecer el lugar exacto en el área precordial, para colocar de forma correcta el sensor (acelerómetro de 3 ejes), ademá...
Presentation
Full-text available
Presentación realizada en BEAMER de LaTeX que muestra los avances en el diseño y construcción de un sistema basado en acelerometría, para la captación y análisis en Matlab, de señales precordiales usando sismocardiografía. De las pruebas realizadas y consultas bibliográficas, se ve que el sismocardiógrafo, como dispositivo biomédico para medir la a...
Conference Paper
Full-text available
Las asignaturas de diseño mecatrónico se centran en el diseño funcional de sistemas autónomos e inteligentes, aplicando diferentes disciplinas tecnológicas (dinámica mecánica, la electrónica de baja y alta potencia, electromecánica, control de movimiento, la óptica, metrología y procesamiento de señales) en un entorno bien equilibrado fortaleciéndo...
Presentation
Full-text available
Presentación realizada en BEAMER de LaTeX donde se expone el proceso de diseño, construcción y puesta en marcha de un robot cartesiano con 3 grados de libertad y la programación de su respectivo software de interfaz con el usuario, este fue realizado en lenguaje Python donde las principales ventajas para su escogencia fueron su versatilidad, portab...
Thesis
Full-text available
This degree project presents development of a accelerometer based system to capture and analyze in Matlab cardiac signals using Seismocardiography (SCG). SCG is an effective non-invasive method for detention of signal in the chest. The project was divided into three phases; the first one was the selection of an accelerometer, data acquisition syste...
Poster
Full-text available
The decontamination of water is a globally issue with the increasing development of production companies that increases the amount of wastes of those companies generate, most of those wastes are thrown into water sources without the proper management or water treatment. The aim of this project is to compare the decontamination methods such as the a...
Poster
Full-text available
SCG is an effective non-invasive method to detention of signal in the chest. The project was divided into three phases; the first one was the selection of an accelerometer, data acquisition system (DAQ) and the communication interface. The second one included the digitalization of the biosignal, and finally, the third phase took place on the SCG di...
Conference Paper
Full-text available
Bioimpedance analysis or bioelectrical impedance provides information related to the degree of hydration and nutrition, which exists in the human body. In recent years it has boomed creating many monitors that provide us with different items, such as, fat mass, fat-free mass, total body water, among others. This article aims to inform the principle...
Poster
Full-text available
Bioimpedance analysis or bioelectrical impedance provides information related to the degree of hydration and nutrition which exists in the human body. In recent years it has boomed creating many monitors that provide us with this information you can provide us with different items, such as, fat mass, lean mass, total body water, among others. This...

Questions

Questions (4)
Question
Hello, firstly thank for your answers.
I want to make an APP in BLUEMIX that must have a database (DB), in the other hand i have to make a webpage using HTML using BLUEMIX. But i don´t know how to syncronize the DB with the DB of the webpage. In other words, use the same DB for both APP and webpage.
Question
I want to build a system that observe cardiac signals from a pacient via a webpage and this webpage should be optimize for Android system, like an Android APP. This signals will be recorded by a arduino set in the chest, and will send via bluetooth the recorded signals to the Android App after to an a database that is connected as same as the webpage.
Someone advised to me that use HTML to build this system, but i want alternatives.
Question
I want to create a global system to comunicate arduino and android app with a database and webpage.
First i got the raw signal from a biomedical device and store in a microsd in the arduino, next i want to send this signal via bluetooth to a android app, next save this signal in a database.
Finally i want to create a webpage that can comunicate with this Database as the android app.
Question
I have been working in the development a device to capture SCG, ICG and ECG simultaneous, and this device will going to send this 3 signals to a APP android via Bluetooth, for save the signals in a Database, but i want that the database be useful for both webpage and android app.
Software to develop a APP android to comunicate between Android and Arduino bluetooth.
Software to develop the webpage.
Software to create the Database, maybe MySQL, but i don´t know if currently is free.
Thank's

Network

Cited By

Projects

Projects (3)
Project
ATCO2 project aims to develop infrastructure for automated collection of air traffic control voice data and implement a framework that allows efficient processing, segmentation, and annotation of this data for Machine Learning (ML) methods. The primary use case is VHF (very high frequency) radio communication and transcription, annotation, and processing of recorded data.
Project
Develop a machine learning environment feed by 3 different techniques of breast cancer, blood test + BMI, Thermography and Electrical Impedance Tomography (EIT). The current stage of the project involve testing different ML techniques like Deep Neuronal Networks (DNN), Self-Normalizing Neural Networks (SNN), Gradient boosting machines (GBM) like lGBM, CatBoost and XGBoost for the stacked database (csv databases), Convolutional Neural Networks (CNN) will be tested for the image recognizing database, finally Recurrent Neural Networks (RNN) for the time series analysis of EIT signals. In the other hand, validation techniques like k-fold cross-validation or Monte Carlo cross-validation will be implemented. After the training phase, a intelligent device will be made for the test the whole system.
Archived project
Development of PI and P Controllers for Speed and Position of a DC Motor