arXiv subjects

Abhishek Kumar

Publications and source records attributed to Abhishek Kumar.

At least 163 records · Page 9Linked to original sources

arXiv2020

In Silico Investigations on the Potential Inhibitors for COVID-19 Protease

A novel strain of coronavirus, namely, COVID-19 has been identified in Wuhan city of China in December 2019. There are no specific therapies available and investigations regarding the treatment of the COVID-19 are still lacking. This prompted us to perform a preliminary in silico study on the COVID-19 protease with anti-malarial compounds in the search of potential inhibitor. We have calculated log P and log S values in addition to molecular docking and PASS predictions. Among the seven studied compounds, mepacrine appears as the potential inhibitor of the COVID-19 followed by chloroquine, hydroxychloroquine and phomarin. Therefore, these anti-malarial drugs may be potential drug candidate for the treatment of this novel coronavirus. A detailed analysis on these inhibitors is currently in progress and clinical studies are invited to investigate their potential medicinal use for the COVID-19.

physics.chem-ph↗

arXiv2020

Weakly Supervised Disentanglement with Guarantees

Learning disentangled representations that correspond to factors of variation in real-world data is critical to interpretable and human-controllable machine learning. Recently, concerns about the viability of learning disentangled representations in a purely unsupervised manner has spurred a shift toward the incorporation of weak supervision. However, there is currently no formalism that identifies when and how weak supervision will guarantee disentanglement. To address this issue, we provide a theoretical framework to assist in analyzing the disentanglement guarantees (or lack thereof) conferred by weak supervision when coupled with learning algorithms based on distribution matching. We empirically verify the guarantees and limitations of several weak supervision methods (restricted labeling, match-pairing, and rank-pairing), demonstrating the predictive power and usefulness of our theoretical framework.

cs.LG↗

arXiv2020

On the Inhibition of COVID-19 Protease by Indian Herbal Plants: An In Silico Investigation

COVID-19 has quickly spread across the globe, becoming a pandemic. This disease has a variable impact in different countries depending on their cultural norms, mitigation efforts and health infrastructure. In India, a majority of people rely upon traditional Indian medicine to treat human maladies due to less-cost, easier availability and without any side-effect. These medicines are made by herbal plants. This study aims to assess the Indian herbal plants in the pursuit of potential COVID-19 inhibitors using in silico approaches. We have considered 18 extracted compounds of 11 different species of these plants. Our calculated lipophilicity, aqueous solubility and binding affinity of the extracted compounds suggest that the inhibition potentials in the order; harsingar > aloe vera > giloy > turmeric > neem > ashwagandha > red onion > tulsi > cannabis > black pepper. On comparing the binding affinity with hydroxychloroquine, we note that the inhibition potentials of the extracts of harsingar, aloe vera and giloy are very promising. Therefore, we believe that these findings will open further possibilities and accelerate the works towards finding an antidote for this malady.

q-bio.OT↗

arXiv2020

Non-Markovian resonance fluorescence

We derive a general formula for the non-Markovian fluorescence spectrum of a multi-level system interacting with a bosonic environment. To this end, we apply linear-response theory to describe the dynamics of a detector monitoring the emission spectrum of a general multi-level system. The resultant emission lineshape function is directly related to the two-time correlation of system observables, which we derive using Nakajima-Zwanzig Generalized Master Equation without assuming a Markov approximation.

cond-mat.mes-hall↗

arXiv2020

Marketplace for AI Models

Artificial intelligence shows promise for solving many practical societal problems in areas such as healthcare and transportation. However, the current mechanisms for AI model diffusion such as Github code repositories, academic project webpages, and commercial AI marketplaces have some limitations; for example, a lack of monetization methods, model traceability, and model auditabilty. In this work, we sketch guidelines for a new AI diffusion method based on a decentralized online marketplace. We consider the technical, economic, and regulatory aspects of such a marketplace including a discussion of solutions for problems in these areas. Finally, we include a comparative analysis of several current AI marketplaces that are already available or in development. We find that most of these marketplaces are centralized commercial marketplaces with relatively few models.

cs.CY↗

arXiv2020

Bayesian Structure Adaptation for Continual Learning

Continual Learning is a learning paradigm where learning systems are trained with sequential or streaming tasks. Two notable directions among the recent advances in continual learning with neural networks are ($i$) variational Bayes based regularization by learning priors from previous tasks, and, ($ii$) learning the structure of deep networks to adapt to new tasks. So far, these two approaches have been orthogonal. We present a novel Bayesian approach to continual learning based on learning the structure of deep neural networks, addressing the shortcomings of both these approaches. The proposed model learns the deep structure for each task by learning which weights to be used, and supports inter-task transfer through the overlapping of different sparse subsets of weights learned by different tasks. Experimental results on supervised and unsupervised benchmarks shows that our model performs comparably or better than recent advances in continual learning setting.

cs.LG↗

arXiv2020

Regularized Autoencoders via Relaxed Injective Probability Flow

Invertible flow-based generative models are an effective method for learning to generate samples, while allowing for tractable likelihood computation and inference. However, the invertibility requirement restricts models to have the same latent dimensionality as the inputs. This imposes significant architectural, memory, and computational costs, making them more challenging to scale than other classes of generative models such as Variational Autoencoders (VAEs). We propose a generative model based on probability flows that does away with the bijectivity requirement on the model and only assumes injectivity. This also provides another perspective on regularized autoencoders (RAEs), with our final objectives resembling RAEs with specific regularizers that are derived by lower bounding the probability flow objective. We empirically demonstrate the promise of the proposed model, improving over VAEs and AEs in terms of sample quality.

cs.LG↗

arXiv2020

Elasticity of Jammed Packings of Sticky Disks

Numerous soft materials jam into an amorphous solid at high packing fraction. This non-equilibrium phase transition is best understood in the context of a model system in which particles repel elastically when they overlap. Recently, however, it was shown that introducing any finite amount of attraction between particles changes the universality class of the transition. The properties of this new ``sticky jamming'' class remain almost entirely unexplored. We use molecular dynamics simulations and scaling analysis to determine the shear modulus, bulk modulus, and coordination of marginal solids close to the sticky jamming point. In each case, the behavior of the system departs sharply and qualitatively from the purely repulsive case.

cond-mat.soft↗

arXiv2020

Trustworthy AI in the Age of Pervasive Computing and Big Data

The era of pervasive computing has resulted in countless devices that continuously monitor users and their environment, generating an abundance of user behavioural data. Such data may support improving the quality of service, but may also lead to adverse usages such as surveillance and advertisement. In parallel, Artificial Intelligence (AI) systems are being applied to sensitive fields such as healthcare, justice, or human resources, raising multiple concerns on the trustworthiness of such systems. Trust in AI systems is thus intrinsically linked to ethics, including the ethics of algorithms, the ethics of data, or the ethics of practice. In this paper, we formalise the requirements of trustworthy AI systems through an ethics perspective. We specifically focus on the aspects that can be integrated into the design and development of AI systems. After discussing the state of research and the remaining challenges, we show how a concrete use-case in smart cities can benefit from these methods.

cs.CY↗

arXiv2019

Mixed baroclinic convection in a cavity

We study the convective patterns that arise in a nearly semi-cylindrical cavity fed in with hot fluid at the upper boundary, bounded by a cold, porous semi-circular boundary at the bottom, and infinitely extended in the third direction. While this configuration is relevant to continuous casting processes that are significantly more complex, we focus on the flow patterns associated with the particular form of mixed convection that arises in it. Linear stability analysis and direct numerical simulations (DNS) are conducted, using the spectral element method to identify observable states. The nature of the bifurcations is determined through Stuart--Landau analysis for completeness. The base flow consists of two counter-rotating rolls driven by the baroclinic imbalance due to the curved isothermal boundary. These are however suppressed by the through-flow, which is found to have a stabilising influence as soon as the Reynolds number $Re$ based on the through-flow exceeds 25. For a sufficiently high Rayleigh number, this base flow is linearly unstable to three different modes, depending on $Re$. For $Re\leq75$, the rolls destabilise through a supercritical bifurcation into a travelling wave. For $100\leq Re \leq 110$, a subcritical bifurcation leads to a standing oscillatory mode, whereas for $Re\geq150$, the unstable mode is non-oscillatory and grows out of a supercritical bifurcation. The direct numerical simulations confirm that in all cases, the dominant mode returned by the linear stability analysis precisely matches the topology and evolution of the flow patterns that arise out of the fully nonlinear dynamics.

physics.flu-dyn↗

arXiv2019

Emotion helps Sentiment: A Multi-task Model for Sentiment and Emotion Analysis

In this paper, we propose a two-layered multi-task attention based neural network that performs sentiment analysis through emotion analysis. The proposed approach is based on Bidirectional Long Short-Term Memory and uses Distributional Thesaurus as a source of external knowledge to improve the sentiment and emotion prediction. The proposed system has two levels of attention to hierarchically build a meaningful representation. We evaluate our system on the benchmark dataset of SemEval 2016 Task 6 and also compare it with the state-of-the-art systems on Stance Sentiment Emotion Corpus. Experimental results show that the proposed system improves the performance of sentiment analysis by 3.2 F-score points on SemEval 2016 Task 6 dataset. Our network also boosts the performance of emotion analysis by 5 F-score points on Stance Sentiment Emotion Corpus.

cs.CL↗

arXiv2019

Customized video filtering on YouTube

Inappropriate and profane content on social media is exponentially increasing and big corporations are becoming more aware of the type of content on which they are advertising and how it may affect their brand reputation. But with a huge surge in content being posted online it becomes seemingly difficult to filter out related videos on which they can run their ads without compromising brand name. Advertising on youtube videos generates a huge amount of revenue for corporations. It becomes increasingly important for such corporations to advertise on only the videos that don't hurt the feelings, community or harmony of the audience at large. In this paper, we propose a system to identify inappropriate content on YouTube and leverage it to perform a first of its kind, large scale, quantitative characterization that reveals some of the risks of YouTube ads consumption on inappropriate videos. Customization of the architecture have also been included to serve different requirements of corporations. Our analysis reveals that YouTube is still plagued by such disturbing videos and its currently deployed countermeasures are ineffective in terms of detecting them in a timely manner. Our framework tries to fill this gap by providing a handy, add on solution to filter the videos and help corporations and companies to push ads on the platform without worrying about the content on which the ads are displayed.

cs.LG↗

arXiv2019

Refined $α$-Divergence Variational Inference via Rejection Sampling

We present an approximate inference method, based on a synergistic combination of Rényi $α$-divergence variational inference (RDVI) and rejection sampling (RS). RDVI is based on minimization of Rényi $α$-divergence $D_α(p||q)$ between the true distribution $p(x)$ and a variational approximation $q(x)$; RS draws samples from a distribution $p(x) = \tilde{p}(x)/Z_{p}$ using a proposal $q(x)$, s.t. $Mq(x) \geq \tilde{p}(x), \forall x$. Our inference method is based on a crucial observation that $D_\infty(p||q)$ equals $\log M(θ)$ where $M(θ)$ is the optimal value of the RS constant for a given proposal $q_θ(x)$. This enables us to develop a \emph{two-stage} hybrid inference algorithm. Stage-1 performs RDVI to learn $q_θ$ by minimizing an estimator of $D_α(p||q)$, and uses the learned $q_θ$ to find an (approximately) optimal $\tilde{M}(θ)$. Stage-2 performs RS using the constant $\tilde{M}(θ)$ to improve the approximate distribution $q_θ$ and obtain a sample-based approximation. We prove that this two-stage method allows us to learn considerably more accurate approximations of the target distribution as compared to RDVI. We demonstrate our method's efficacy via several experiments on synthetic and real datasets.

cs.LG↗

arXiv2019

Alpha Particle X-Ray Spectrometer (APXS) On-board Chandrayaan-2 Rover -- Pragyan

Alpha Particle X-ray Spectrometer (APXS) is one of the two scientific experiments on Chandrayaan-2 rover named as Pragyan. The primary scientific objective of APXS is to determine the elemental composition of the lunar surface in the surrounding regions of the landing site. This will be achieved by employing the technique of X-ray fluorescence spectroscopy using in-situ excitation source Cm-244 emitting both X-rays and alpha particles. These radiations excite characteristic X-rays of the elements by the processes of particle induced X-ray emission (PIXE) and X-ray fluorescence (XRF). The characteristic X-rays are detected by the state-of-the-art X-ray detector known as Silicon Drift Detector (SDD), which provides high energy resolution as well as high efficiency in the energy range of 1 to 25 keV. This enables APXS to detect all major rock forming elements such as, Na, Mg, Al, Si, Ca, Ti and Fe. The Flight Model (FM) of the APXS payload has been completed and tested for various instrument parameters. The APXS provides energy resolution of 135 eV at 5.9 keV for the detector operating temperature of about -35 deg C. The design details and the performance measurement of APXS are presented in this paper.

astro-ph.IM↗

arXiv2019

Turbulence in a stably stratified fluid: Onset of global anisotropy as a function of the Richardson number

It is necessary to introduce an external forcing to induce turbulence in a stably stratified fluid. The Heisenberg eddy viscosity technique should in this case suffice to calculate a space-time averaged quantity like the global anisotropy parameter as a function of the Richardson number. We find analytically that the anisotropy increases linearly with the Richardson number, with a small quadratic correction. A numerical simulation of the complete equations shows the linear behaviour.

physics.flu-dyn↗

arXiv2019

Transient flows and reorientations of large-scale convection in a cubic cell

The transient processes of a turbulent large-scale convective circulation (LSC) in a cubic cell are investigated using large-eddy simulations for Rayleigh number $\Ray=10^8$ and Prandtl number $\Pran=0.7$. For the first time, we have explicitly shown that LSC is accompanied by large-scale azimuthal flows with non-zero total angular momentum. It is also shown that solid-body rotation of the entire fluid is not realized. It is found that correlation between rotation of LSC plane and the mean azimuthal motion is high during quasiperiodic oscillations of LSC near the diagonal plane and relatively weak during LSC reorientations. We propose a new plausible scenario for the reorientations of the LSC in a cube that does not involve a mean azimuthal flow. Instead of a single-roll, we introduce the superposition of a pair of large-scale orthogonal quasi-two-dimensional (Q2D) rolls and the reorientation of the LSC occurs as a result of the cessation of one of the Q2D rolls. This scenario is consistent with all known experimental and numerical data.

physics.flu-dyn↗

arXiv2019

Higher-order Floquet topological phases with corner and bulk bound states

We report the theoretical discovery and characterization of higher-order Floquet topological phases dynamically generated in a periodically driven system with mirror symmetries. We demonstrate numerically and analytically that these phases support lower-dimensional Floquet bound states, such as corner Floquet bound states at the intersection of edges of a two-dimensional system, protected by the nonequilibrium higher-order topology induced by the periodic drive. We characterize higher-order Floquet topologies of the bulk Floquet Hamiltonian using mirror-graded Floquet topological invariants. This allows for the characterization of a new class of higher-order "anomalous" Floquet topological phase, where the corners of the open system host Floquet bound states with the same as well as with double the period of the drive. Moreover, we show that bulk vortex structures can be dynamically generated by a drive that is spatially inhomogeneous. We show these bulk vortices can host multiple Floquet bound states. This "stirring drive protocol" leverages a connection between higher-order topologies and previously studied fractionally charged, bulk topological defects. Our work establishes Floquet engineering of higher-order topological phases and bulk defects beyond equilibrium classification and offers a versatile tool for dynamical generation and control of topologically protected Floquet corner and bulk bound states.

cond-mat.mes-hall↗

arXiv2019

Sweeping effect and Taylor's hypothesis via correlation function

We demonstrate the sweeping effect in turbulence using numerical simulations of hydrodynamic turbulence without a mean velocity. The velocity correlation function, C(k, τ ) decays with time due to the eddy viscosity. In addition, C(k, τ ) shows oscillations due to the sweeping effect by "random mean velocity field" U_0. We also perform numerical simulation with a mean velocity U_0 = 10 for which C(k, τ ) exhibits damped oscillations with the frequency of |U_0|k and decay time scale corresponding to the U_0 = 0 case. For U_0 = 10z, the phase of C(k, τ ) show the sweeping effect, but it is overshadowed by oscillations caused by U_0 . We also demonstrate that for U0 = 0 and 10z, the frequency spectra of the velocity fields measured by real-space probes are respectively f^{-2} and f^{-5/3} these spectra are related to the Lagrangian and Eulerian space-time correlations respectively. respectively.

physics.flu-dyn↗