Pablo Zegers

Computational Intelligence Challenges and Applications on Large-Scale Astronomical Time Series Databases

Time-domain astronomy (TDA) is facing a paradigm shift caused by the exponential growth of the sample size, data complexity and data generation rates of new astronomical sky surveys. For example, the Large Synoptic Survey Telescope (LSST), which will begin operations in northern Chile in 2022, will generate a nearly 150 Petabyte imaging dataset of the southern hemisphere sky. The LSST will stream data at rates of 2 Terabytes per hour, effectively capturing an unprecedented movie of the sky. The LSST is expected not only to improve our understanding of time-varying astrophysical objects, but also to reveal a plethora of yet unknown faint and fast-varying phenomena. To cope with a change of paradigm to data-driven astronomy, the fields of astroinformatics and astrostatistics have been created recently. The new data-oriented paradigms for astronomy combine statistics, data mining, knowledge discovery, machine learning and computational intelligence, in order to provide the automated and robust methods needed for the rapid detection and classification of known astrophysical objects as well as the unsupervised characterization of novel phenomena. In this article we present an overview of machine learning and computational intelligence applications to TDA. Future big data challenges and new lines of research in TDA, focusing on the LSST, are identified and discussed from the viewpoint of computational intelligence/machine learning. Interdisciplinary collaboration will be required to cope with the challenges posed by the deluge of astronomical data coming from the LSST.

An Information Theoretic Algorithm for Finding Periodicities in Stellar Light Curves

We propose a new information theoretic metric for finding periodicities in stellar light curves. Light curves are astronomical time series of brightness over time, and are characterized as being noisy and unevenly sampled. The proposed metric combines correntropy (generalized correlation) with a periodic kernel to measure similarity among samples separated by a given period. The new metric provides a periodogram, called Correntropy Kernelized Periodogram (CKP), whose peaks are associated with the fundamental frequencies present in the data. The CKP does not require any resampling, slotting or folding scheme as it is computed directly from the available samples. CKP is the main part of a fully-automated pipeline for periodic light curve discrimination to be used in astronomical survey databases. We show that the CKP method outperformed the slotted correntropy, and conventional methods used in astronomy for periodicity discrimination and period estimation tasks, using a set of light curves drawn from the MACHO survey. The proposed metric achieved 97.2% of true positives with 0% of false positives at the confidence level of 99% for the periodicity discrimination task; and 88% of hits with 11.6% of multiples and 0.4% of misses in the period estimation task.

Period Estimation in Astronomical Time Series Using Slotted Correntropy

In this letter, we propose a method for period estimation in light curves from periodic variable stars using correntropy. Light curves are astronomical time series of stellar brightness over time, and are characterized as being noisy and unevenly sampled. We propose to use slotted time lags in order to estimate correntropy directly from irregularly sampled time series. A new information theoretic metric is proposed for discriminating among the peaks of the correntropy spectral density. The slotted correntropy method outperformed slotted correlation, string length, VarTools (Lomb-Scargle periodogram and Analysis of Variance), and SigSpec applications on a set of light curves drawn from the MACHO survey.

People Counting in Videos by Fusing Temporal Cues from Spatial Context-Aware Convolutional Neural Networks

We present an efficient method for people counting in video sequences from fixed cameras by utilising the responses of spatially context-aware convolutional neural networks (CNN) in the temporal domain. For stationary cameras, the background information remains fairly static, while foreground characteristics, such as size and orientation may depend on their image location, thus the use of whole frames for training a CNN improves the differentiation between background and foreground pixels. Foreground density representing the presence of people in the environment can then be associated with people counts. Moreover the fusion, of the responses of count estimations, in the temporal domain, can further enhance the accuracy of the final count. Our methodology was tested using the publicly available Mall dataset and achieved a mean deviation error of 0.091.

A novel, fully automated pipeline for period estimation in the EROS 2 data set

We present a new method to discriminate periodic from nonperiodic irregularly sampled light curves. We introduce a periodic kernel and maximize a similarity measure derived from information theory to estimate the periods and a discriminator factor. We tested the method on a data set containing 100,000 synthetic periodic and nonperiodic light curves with various periods, amplitudes, and shapes generated using a multivariate generative model. We correctly identified periodic and nonperiodic light curves with a completeness of ~90% and a precision of ~95%, for light curves with a signal-to-noise ratio (S/N) larger than 0.5. We characterize the efficiency and reliability of the model using these synthetic light curves and apply the method on the EROS-2 data set. A crucial consideration is the speed at which the method can be executed. Using a hierarchical search and some simplification on the parameter search, we were able to analyze 32.8 million light curves in ~18 hr on a cluster of GPGPUs. Using the sensitivity analysis on the synthetic data set, we infer that 0.42% of the sources in the LMC and 0.61% of the sources in the SMC show periodic behavior. The training set, catalogs, and source code are all available at http://timemachine.iic.harvard.edu.

Exploiting deep residual networks for human action recognition from skeletal data

The computer vision community is currently focusing on solving action recognition problems in real videos, which contain thousands of samples with many challenges. In this process, Deep Convolutional Neural Networks (D-CNNs) have played a significant role in advancing the state-of-the-art in various vision-based action recognition systems. Recently, the introduction of residual connections in conjunction with a more traditional CNN model in a single architecture called Residual Network (ResNet) has shown impressive performance and great potential for image recognition tasks. In this paper, we investigate and apply deep ResNets for human action recognition using skeletal data provided by depth sensors. Firstly, the 3D coordinates of the human body joints carried in skeleton sequences are transformed into image-based representations and stored as RGB images. These color images are able to capture the spatial-temporal evolutions of 3D motions from skeleton sequences and can be efficiently learned by D-CNNs. We then propose a novel deep learning architecture based on ResNets to learn features from obtained color-based representations and classify them into action classes. The proposed method is evaluated on three challenging benchmark datasets including MSR Action 3D, KARD, and NTU-RGB+D datasets. Experimental results demonstrate that our method achieves state-of-the-art performance for all these benchmarks whilst requiring less computation resource. In particular, the proposed method surpasses previous approaches by a significant margin of 3.4% on MSR Action 3D dataset, 0.67% on KARD dataset, and 2.5% on NTU-RGB+D dataset.

Exploiting Spatio-Temporal Structure With Recurrent Winner-Take-All Networks

We propose a convolutional recurrent neural network, with Winner-Take-All dropout for high dimensional unsupervised feature learning in multi-dimensional time series. We apply the proposedmethod for object recognition with temporal context in videos and obtain better results than comparable methods in the literature, including the Deep Predictive Coding Networks previously proposed by Chalasani and Principe.Our contributions can be summarized as a scalable reinterpretation of the Deep Predictive Coding Networks trained end-to-end with backpropagation through time, an extension of the previously proposed Winner-Take-All Autoencoders to sequences in time, and a new technique for initializing and regularizing convolutional-recurrent neural networks.

Learning and recognizing human action from skeleton movement with deep residual neural networks

Automatic human action recognition is indispensable for almost artificial intelligent systems such as video surveillance, human-computer interfaces, video retrieval, etc. Despite a lot of progress, recognizing actions in an unknown video is still a challenging task in computer vision. Recently, deep learning algorithms have proved its great potential in many vision-related recognition tasks. In this paper, we propose the use of Deep Residual Neural Networks (ResNets) to learn and recognize human action from skeleton data provided by Kinect sensor. Firstly, the body joint coordinates are transformed into 3D-arrays and saved in RGB images space. Five different deep learning models based on ResNet have been designed to extract image features and classify them into classes. Experiments are conducted on two public video datasets for human action recognition containing various challenges. The results show that our method achieves the state-of-the-art performance comparing with existing approaches.

Uncalibrated Neuro-Visual Servoing Control for Multiple Robot Arms

Diverse image-based tracking schemes for a robot moving in free motion have been proposed and experimentally validated. However, few visual servoing schemes have addressed the tracking of the desired trajectory and the contact forces for multiple robot arms. The main difficulty stems from the fact that camera information cannot be used to drive force trajectories. Recognizing this fact, a unique error manifold that includes position-velocity and force errors in orthogonal complements is proposed. A synergistic scheme that fuses camera, encoder and force sensor signals into a unique error manifold allows proposing a control system which guarantees exponential tracking errors under parametric uncertainty. Additionally a small neural network driven by a second order sliding mode surface is derived to compensate robot dynamics. Residual errors that arise because of the finite size of the neural network are compensated via an orthogonalized second order sliding mode. The performance of the proposed scheme, in two significant applications of the multiple robot arms, is validated through numerical simulations.

Bounded-time system identification under neuro-sliding training

A neural network training method for identification in bounded time of nonlinear systems is presented in this paper. A sliding mode surface drives the adalines, perceptrons and multilayer perceptrons so as to a new second order sliding mode is enforced for all time. This neural network-based sliding mode enforces an invariant differential manifold, with a time-varying feedback gain to give rise to finite-time convergence, consequently, the chattering free sliding mode allows identification of the underlying system in finite-time, with zero error. Convergence characteristics of the algorithm are proven with Lyapunov stability theory and concepts drawn from variable structure systems. Numerical simulations for a full nonlinear nonlinear robot arm, subject to noise, show the validity of the proposed approach.