arXiv subjects

Aritra Ghosh

Publications and source records attributed to Aritra Ghosh.

94 records · Page 6Linked to original sources

arXiv2018

Probing the Jets of Blazars Using the Temporal Symmetry of Their Multi-Wavelength Outbursts

We compare the rise and decay timescales of $\sim$200 long-term ($\sim$weeks-months) GeV and R-band outbursts and $\sim$25 short-term ($\sim$hr-day) GeV flares in a sample of 10 blazars using light curves from the Fermi-LAT and the Yale/SMARTS monitoring project. We find that most of the long-term outbursts are symmetric, indicating that the observed variability is dominated by the crossing timescale of a disturbance, e.g., a shock. A larger fraction of short-term flares are asymmetric with an approximately equal fraction of longer and shorter decay than rise timescale. We employ the MUlti-ZOne Radiation Feedback (MUZORF) model to interpret the above results. We find that the outbursts with slow rise times indicate a gradual acceleration of the particles to GeV energy. A change in the bulk Lorentz factor of the plasma or the width of the shocked region can lead to an increase of the cooling time causing a faster rise than decay time. Parameters such as the luminosity or the distance of the broad line region (BLR) affects the cooling time strongly if a single emission mechanism, e.g., external Compton scattering of BLR photons is considered but may not if other mechanisms, e.g., synchrotron self-compton and external Compton scattering of the torus photon are included. This work carries out a systematic study of the symmetry of flares, which can be used to estimate relevant geometric and physical parameters of blazar jets in the context of the MUZORF model.

astro-ph.GA↗

arXiv2017

Robust Loss Functions under Label Noise for Deep Neural Networks

In many applications of classifier learning, training data suffers from label noise. Deep networks are learned using huge training data where the problem of noisy labels is particularly relevant. The current techniques proposed for learning deep networks under label noise focus on modifying the network architecture and on algorithms for estimating true labels from noisy labels. An alternate approach would be to look for loss functions that are inherently noise-tolerant. For binary classification there exist theoretical results on loss functions that are robust to label noise. In this paper, we provide some sufficient conditions on a loss function so that risk minimization under that loss function would be inherently tolerant to label noise for multiclass classification problems. These results generalize the existing results on noise-tolerant loss functions for binary classification. We study some of the widely used loss functions in deep networks and show that the loss function based on mean absolute value of error is inherently robust to label noise. Thus standard back propagation is enough to learn the true classifier even under label noise. Through experiments, we illustrate the robustness of risk minimization with such loss functions for learning neural networks.

stat.ML↗

arXiv2016

On the Robustness of Decision Tree Learning under Label Noise

In most practical problems of classifier learning, the training data suffers from the label noise. Hence, it is important to understand how robust is a learning algorithm to such label noise. This paper presents some theoretical analysis to show that many popular decision tree algorithms are robust to symmetric label noise under large sample size. We also present some sample complexity results which provide some bounds on the sample size for the robustness to hold with a high probability. Through extensive simulations we illustrate this robustness.

cs.LG↗

arXiv2015

Making Risk Minimization Tolerant to Label Noise

In many applications, the training data, from which one needs to learn a classifier, is corrupted with label noise. Many standard algorithms such as SVM perform poorly in presence of label noise. In this paper we investigate the robustness of risk minimization to label noise. We prove a sufficient condition on a loss function for the risk minimization under that loss to be tolerant to uniform label noise. We show that the $0-1$ loss, sigmoid loss, ramp loss and probit loss satisfy this condition though none of the standard convex loss functions satisfy it. We also prove that, by choosing a sufficiently large value of a parameter in the loss function, the sigmoid loss, ramp loss and probit loss can be made tolerant to non-uniform label noise also if we can assume the classes to be separable under noise-free data distribution. Through extensive empirical studies, we show that risk minimization under the $0-1$ loss, the sigmoid loss and the ramp loss has much better robustness to label noise when compared to the SVM algorithm.

cs.LG↗