Faculty Publications

Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736

Publications by NITK Faculty

Browse

Author: search.filters.author.Choudhury, S.

Search Results

Now showing 1 - 5 of 5
  • No Thumbnail Available
    Item
    Comparative Study of Pruning Techniques in Recurrent Neural Networks
    (Springer Science and Business Media Deutschland GmbH, 2023) Choudhury, S.; Rout, A.K.; Pragnesh, T.; Mohan, B.R.
    In recent years, there has been a drastic development in the field of neural networks. They have evolved from simple feed-forward neural networks to more complex neural networks such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs). CNNs are used for tasks such as image recognition where the sequence is not essential, while RNNs are useful when order is important such as machine translation. By increasing the number of layers in the network, we can improve the performance of the neural network (Alford et al. in Pruned and structurally sparse neural networks, 2018 [1]). However, this will also increase the complexity of the network, and also training will require more power and time. By introducing sparsity in the architecture of the neural network, we can tackle this problem. Pruning is one of the processes through which a neural network can be made sparse (Zhu and Gupta in To prune, or not to prune: exploring the efficacy of pruning for model compression, 2017 [2]). Sparse RNNs can be easily implemented on mobile devices and resource-constraint servers (Wen et al. in Learning intrinsic sparse structures within long short-term memory, 2017 [3]). We investigate the following methods to induce sparsity in RNNs: RNN pruning and automated gradual pruning. We also investigate how the pruning techniques impact the model’s performance and provide a detailed comparison between the two techniques. We also experiment by pruning input-to-hidden and hidden-to-hidden weights. Based on the results of pruning experiments, we conclude that it is possible to reduce the complexity of RNNs by more than 80%. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
  • No Thumbnail Available
    Item
    Entanglement in interacting quenched two-body coupled oscillator system
    (American Physical Society, 2022) Choudhury, S.; Gharat, R.M.; Mandal, S.; Pandey, N.; Roy, A.; Sarker, P.
    In this work, we explore the effects of a quantum quench on the entanglement measures of a two-body coupled oscillator system having quartic interaction. We use the invariant operator method, under a perturbative framework, for computing the ground state of this system. We give the analytical expressions for the total and reduced density matrix of the system having non-Gaussian, quartic interaction terms. Using this reduced density matrix, we show the analytical calculation of two entanglement measures viz., Von Neumann entanglement entropy using replica trick and Renyi entanglement entropy. Further, we give a numerical estimate of these entanglement measures with respect to the dimensionless parameter (t/δt) and show its behavior in the three regimes, i.e., late time behavior, around the quench point and the early time behavior. We comment on the variation of these entanglement measures for different orders of coupling strength. The variation of Renyi entropy of different orders has also been discussed. © 2022 authors. Published by the American Physical Society.
  • No Thumbnail Available
    Item
    Circuit Complexity in Z2 EEFT
    (MDPI, 2023) Adhikari, K.; Choudhury, S.; Kumar, S.; Mandal, S.; Pandey, N.; Roy, A.; Sarkar, S.; Sarker, P.; Shariff, S.S.
    Motivated by recent studies of circuit complexity in weakly interacting scalar field theory, we explore the computation of circuit complexity in (Formula presented.) Even Effective Field Theories ((Formula presented.) EEFTs). We consider a massive free field theory with higher-order Wilsonian operators such as (Formula presented.), (Formula presented.), and (Formula presented.) To facilitate our computation, we regularize the theory by putting it on a lattice. First, we consider a simple case of two oscillators and later generalize the results to N oscillators. This study was carried out for nearly Gaussian states. In our computation, the reference state is an approximately Gaussian unentangled state, and the corresponding target state, calculated from our theory, is an approximately Gaussian entangled state. We compute the complexity using the geometric approach developed by Nielsen, parameterizing the path-ordered unitary transformation and minimizing the geodesic in the space of unitaries. The contribution of higher-order operators to the circuit complexity in our theory is discussed. We also explore the dependency of complexity on other parameters in our theory for various cases. © 2022 by the authors.
  • No Thumbnail Available
    Item
    Circuit Complexity in Interacting Quenched Quantum Field Theory
    (MDPI, 2023) Choudhury, S.; Gharat, R.M.; Mandal, S.; Pandey, N.
    In this work, we explore the effects of quantum quenching on the circuit complexity of quenched quantum field theory with weakly coupled quartic interactions. We use the invariant operator method under a perturbative framework to compute the ground state of this system. We give the analytical expressions for specific reference and target states using the ground state of the system. Using a particular cost functional, we show the analytical computation of circuit complexity for the quenched and interacting field theory. Furthermore, we give a numerical estimate of circuit complexity with respect to the quench rate, (Formula presented.), for two coupled oscillators. The parametric variation in the unambiguous contribution of the circuit complexity for an arbitrary number of oscillators has been studied with respect to the dimensionless parameter (Formula presented.)). We comment on the variation in the circuit complexity for different values of coupling strength, different numbers of oscillators and even in different dimensions. © 2023 by the authors.
  • No Thumbnail Available
    Item
    Schwinger–Keldysh Path Integral Formalism for a Quenched Quantum Inverted Oscillator
    (Multidisciplinary Digital Publishing Institute (MDPI), 2024) Choudhury, S.; Dey, S.; Gharat, R.M.; Mandal, S.; Pandey, N.
    In this work, we study the time-dependent behavior of quantum correlations of a system of an inverted oscillator governed by out-of-equilibrium dynamics using the well-known Schwinger–Keldysh formalism in the presence of quantum mechanical quench. Considering a generalized structure of a time-dependent Hamiltonian for an inverted oscillator system, we use the invariant operator method to obtain its eigenstate and continuous energy eigenvalues. Using the expression for the eigenstate, we further derive the most general expression for the generating function as well as the out-of-time-ordered correlators (OTOCs) for the given system using this formalism. Further, considering the time-dependent coupling and frequency of the quantum inverted oscillator characterized by quench parameters, we comment on the dynamical behavior, specifically the early, intermediate and late time-dependent features of the OTOC for the quenched quantum inverted oscillator. Next, we study a specific case, where the system of an inverted oscillator exhibits chaotic behavior by computing the quantum Lyapunov exponent from the time-dependent behavior of OTOCs in the presence of the given quench profile. © 2024 by the authors.