Skip to main content

SISTEMAS DINAMICOS COMPLEJOS Y HERRAMIENTAS AVANZADAS DE ANALISIS DE DATOS

PGC2018-099443-B-I00

•
Funding agency name Agencia Estatal de Investigación
Funding agency acronym AEI
Program Programa Estatal de Generación de Conocimiento y Fortalecimiento Científico y Tecnológico del Sistema de I+D+i
Subprogram Subprograma Estatal de Generación de Conocimiento
Call Proyectos de I+D (Excelencia / Generación de Conocimiento)
Call year 2018
State Plan 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Beneficiary centre UNIVERSITAT POLITECNICA DE CATALUNYA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 6
Encontrada(s) 1 página(s)

UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

New algorithms to characterize and classify ophthalmic images

  • Amil, Pablo
  • Masoller Alonso, Cristina|||0000-0003-0768-2019
Unprecedented advances in machine learning have led to a variety of algorithms for the remote evaluation of biomedical images, allowing for cost-effective early detection of diseases. In particular, a lot of efforts are focused on the development of reliable image analysis tools for the early diagnosis of eye diseases. Here we present several new methods for ophthalmic image analysis. We propose a machine learning algorithm for ordering images of the anterior chamber (optical coherence tomography, OCT), which extracts features that discriminate between healthy subjects and patients with angle-closure. We also present an algorithm to detect the OCT images that contain artifacts, and we show that removing these images from the data base improves the performance of the ordering algorithm. Finally, we present algorithms for the analysis of retina fundus images, which are able to segment the vessel network in the retina and extract features from the topological tree-like network structure. We show that these features discriminate between healthy subjects and those with glaucoma or diabetic retinopathy., P. A. acknowledge support of the Marie Sklodowska-Curie Innovative Training Network Advanced BiomEdical OPTICAL Imaging and Data Analysis (BE-OPTICAL, H2020-675512, http://beoptical.eu). C. M. acknowledges support from the Spanish Ministerio de Ciencia, Innovación y Universidades (PGC2018-099443-B-I00) and the program ICREA ACADEMIA funded by the Generalitat de Catalunya., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Forecasting the amplitude of high-intensity chaotic laser pulses

  • Amil, Pablo
  • Cornelles Soriano, Miguel
  • Masoller Alonso, Cristina|||0000-0003-0768-2019
Forecasting the dynamics of chaotic systems from the analysis of their output signals is a challenging problem with applications in most fields of modern science. In this work, we use a laser model to compare the performance of several machine learning algorithms for forecasting the amplitude of upcoming emitted chaotic pulses. We simulate the dynamics of an optically injected semiconductor laser that presents a rich variety of dynamical regimes when changing the parameters. We focus on a particular regime where the intensity shows a chaotic pulsing dynamics, and occasionally an ultra-high pulse, reminiscent of a rogue wave, is emitted. Our goal is to predict the amplitude (height) of the next pulse, knowing the amplitude of the three preceding pulses. We compare the performance of several machine learning methods, namely neural networks, support vector machine, nearest neighbors and reservoir computing. We analyze how their performance depends on the length of the time-series used for training., P. A. acknowledge support of the Marie Sklodowska-Curie Innovative Training Network Advanced BiomEdical OPTICAL Imaging and Data Analysis (BE-OPTICAL, H2020-675512, http://beoptical.eu). C. M. acknowledges support from the Spanish Ministerio de Ciencia, Innovación y Universidades (PGC2018-099443-B-I00) and ICREA ACADEMIA. M. C. S. was supported through a \Ramon y Cajal" Fellowship (RYC-2015-18140). This work is the result of a collaboration established within the Ibersinc network of excellence (FIS2017-90782-REDT), Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Experimental study of the neuron-like dynamics of a semiconductor laser with optical feedback and current modulation

  • Tiana Alsina, Jordi|||0000-0001-8359-9378
  • Quintero Quiroz, Carlos Alberto|||0000-0003-3520-2326
  • Masoller Alonso, Cristina|||0000-0003-0768-2019
Excitable lasers that mimic neuronal activity can be building blocks of ultra-fast neuro-inspired information processing systems, which can revolutionize the fields of optical signal processing, optical computing and artificial intelligence. To implement such photonic neurons, we must first identify low-cost and energy-efficient lasers whose excitable dynamics mimics neuronal dynamics. Here we conduct an experimental study of the dynamics of a diode laser with optical feedback from an external cavity. We focus on the response of the laser when is under weak periodic perturbations that are implemented via direct modulation of the laser pump current. We consider sinusoidal and pulsed waveforms. The dynamics of the laser intensity is compared with the dynamics of the membrane potential of a neuron, which is simulated with the FitzHugh-Nagumo (FHN) model. The analysis of the statistics of the time intervals between dropouts of the laser intensity (optical spikes) and of the time intervals between the neuron's action potentials (or spikes) unveils similarities, and also differences. The analysis of the spike rate reveals a variation with the amplitude and with the frequency of the external signal that is similar for both, the laser and the FHN neuron. Therefore, in terms of the spike rate, the laser response to a weak periodic signal mimics the response of a FHN neuron, and thus diode lasers with optical feedback can be used to implement photonic neurons. A drawback of this implementation is the length of the feedback cavity, which prevents on-chip integration., This work was supported in part by the Spanish Ministerio de Ciencia, Innovación y Universidades (PGC2018- 099443-B-I00). C. M. also acknowledges support of the ICREA ACADEMIA, Generalitat de Catalunya., Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

New algorithms to characterize and classify ophthalmic images

  • Amil, Pablo
  • Masoller Alonso, Cristina
Unprecedented advances in machine learning have led to a variety of algorithms for the remote evaluation of biomedical images, allowing for cost-effective early detection of diseases. In particular, a lot of efforts are focused on the development of reliable image analysis tools for the early diagnosis of eye diseases. Here we present several new methods for ophthalmic image analysis. We propose a machine learning algorithm for ordering images of the anterior chamber (optical coherence tomography, OCT), which extracts features that discriminate between healthy subjects and patients with angle-closure. We also present an algorithm to detect the OCT images that contain artifacts, and we show that removing these images from the data base improves the performance of the ordering algorithm. Finally, we present algorithms for the analysis of retina fundus images, which are able to segment the vessel network in the retina and extract features from the topological tree-like network structure. We show that these features discriminate between healthy subjects and those with glaucoma or diabetic retinopathy., P. A. acknowledge support of the Marie Sklodowska-Curie Innovative Training Network Advanced BiomEdical OPTICAL Imaging and Data Analysis (BE-OPTICAL, H2020-675512, http://beoptical.eu). C. M. acknowledges support from the Spanish Ministerio de Ciencia, Innovación y Universidades (PGC2018-099443-B-I00) and the program ICREA ACADEMIA funded by the Generalitat de Catalunya., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Forecasting the amplitude of high-intensity chaotic laser pulses

  • Amil, Pablo
  • Cornelles Soriano, Miguel
  • Masoller Alonso, Cristina
Forecasting the dynamics of chaotic systems from the analysis of their output signals is a challenging problem with applications in most fields of modern science. In this work, we use a laser model to compare the performance of several machine learning algorithms for forecasting the amplitude of upcoming emitted chaotic pulses. We simulate the dynamics of an optically injected semiconductor laser that presents a rich variety of dynamical regimes when changing the parameters. We focus on a particular regime where the intensity shows a chaotic pulsing dynamics, and occasionally an ultra-high pulse, reminiscent of a rogue wave, is emitted. Our goal is to predict the amplitude (height) of the next pulse, knowing the amplitude of the three preceding pulses. We compare the performance of several machine learning methods, namely neural networks, support vector machine, nearest neighbors and reservoir computing. We analyze how their performance depends on the length of the time-series used for training., P. A. acknowledge support of the Marie Sklodowska-Curie Innovative Training Network Advanced BiomEdical OPTICAL Imaging and Data Analysis (BE-OPTICAL, H2020-675512, http://beoptical.eu). C. M. acknowledges support from the Spanish Ministerio de Ciencia, Innovación y Universidades (PGC2018-099443-B-I00) and ICREA ACADEMIA. M. C. S. was supported through a \Ramon y Cajal" Fellowship (RYC-2015-18140). This work is the result of a collaboration established within the Ibersinc network of excellence (FIS2017-90782-REDT), Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Experimental study of the neuron-like dynamics of a semiconductor laser with optical feedback and current modulation

  • Tiana Alsina, Jordi
  • Quintero Quiroz, Carlos Alberto
  • Masoller Alonso, Cristina
Excitable lasers that mimic neuronal activity can be building blocks of ultra-fast neuro-inspired information processing systems, which can revolutionize the fields of optical signal processing, optical computing and artificial intelligence. To implement such photonic neurons, we must first identify low-cost and energy-efficient lasers whose excitable dynamics mimics neuronal dynamics. Here we conduct an experimental study of the dynamics of a diode laser with optical feedback from an external cavity. We focus on the response of the laser when is under weak periodic perturbations that are implemented via direct modulation of the laser pump current. We consider sinusoidal and pulsed waveforms. The dynamics of the laser intensity is compared with the dynamics of the membrane potential of a neuron, which is simulated with the FitzHugh-Nagumo (FHN) model. The analysis of the statistics of the time intervals between dropouts of the laser intensity (optical spikes) and of the time intervals between the neuron's action potentials (or spikes) unveils similarities, and also differences. The analysis of the spike rate reveals a variation with the amplitude and with the frequency of the external signal that is similar for both, the laser and the FHN neuron. Therefore, in terms of the spike rate, the laser response to a weak periodic signal mimics the response of a FHN neuron, and thus diode lasers with optical feedback can be used to implement photonic neurons. A drawback of this implementation is the length of the feedback cavity, which prevents on-chip integration., This work was supported in part by the Spanish Ministerio de Ciencia, Innovación y Universidades (PGC2018- 099443-B-I00). C. M. also acknowledges support of the ICREA ACADEMIA, Generalitat de Catalunya., Peer Reviewed, Postprint (author's final draft)