Kundan Kumar

Learn More
Natural image modeling is a landmark challenge of unsupervised learning. Variational Autoencoders (VAEs) learn a useful latent representation and model global structure well but have difficulty capturing small details. PixelCNN models details very well, but lacks a latent code and is difficult to scale for capturing large structures. We present PixelVAE, a(More)
In this paper we propose a novel model for unconditional audio generation based on generating one audio sample at a time. We show that our model, which profits from combining memory-less modules, namely autoregressive multilayer perceptrons, and stateful recurrent neural networks in a hierarchical structure is able to capture underlying sources of(More)
The major hindrance in the development and sustainability of aquaculture industry is the occurrence of various diseases in the farming systems. Today, preventive and management measures are central concern to overcome such outbreak of diseases. Immunostimulants are considered as an effective tool for enhancing immune status of cultured organisms. Among(More)
A cellulase-free xylanase production by Thermomyces lanuginosus SSBP using bagasse pulp was examined under submerged (SmC) and solid-state cultivation (SSC). Higher level of xylanase activity (19,320 ± 37 U g−1 dried carbon source) was obtained in SSC cultures than in SmC (1,772 ± 15 U g−1 dried carbon source) after 120 h with 10% inoculum. The biobleaching(More)
Plants, as sessile organisms experience various abiotic stresses, which pose serious threat to crop production. Plants adapt to environmental stress by modulating their growth and development along with the various physiological and biochemical changes. This phenotypic plasticity is driven by the activation of specific genes encoding signal transduction,(More)
We present Char2Wav, an end-to-end model for speech synthesis. Char2Wav has two components: a reader and a neural vocoder. The reader is an encoderdecoder model with attention. The encoder is a bidirectional recurrent neural network that accepts text or phonemes as inputs, while the decoder is a recurrent neural network (RNN) with attention that produces(More)
Fusarium head blight (FHB) and deoxynivalenol (DON) mycotoxin produced by Fusarium graminearum reduce barley yield and quality worldwide. Hundreds of quantitative trait loci (QTLs) have been identified in wheat and barley but their functions are largely unknown. Metabolic profiling was applied to better understand the mechanisms of resistance and to(More)
Hematological disorders are mostly identified based on characterization of blood parameters i.e. erythrocytes, leukocytes and platelets. Microscopic examination of leukocytes in blood slides is the most frequent laboratory investigation performed for malignancy detection. Hematological examination of blood is an indispensable technique still today and(More)
The actual traffic data collected on various applications specific on-chip networks exposed that the network traffic is self-similar in nature. In this work, modeling of self-similar traffic by aggregation of a large number of on-off Pareto sources has been discussed. We have developed a cycle accurate network simulator for evaluating the performance of(More)
External duodenal fistulae from sutured duodenal ulcer perforation sites are difficult to manage and most patients succumb to septicemia and undernutrition. This is due to failure of closure of the perforation site in the duodenum. Most techniques described in the past to facilitate closure have failed to give satisfactory results. We have devised a new(More)