Tejas Jayashankar
YOU?
Author Swipe
View article: Score-of-Mixture Training: Training One-Step Generative Models Made Simple via Score Estimation of Mixture Distributions
Score-of-Mixture Training: Training One-Step Generative Models Made Simple via Score Estimation of Mixture Distributions Open
We propose Score-of-Mixture Training (SMT), a novel framework for training one-step generative models by minimizing a class of divergences called the $α$-skew Jensen--Shannon divergence. At its core, SMT estimates the score of mixture dist…
View article: Correction to “RF Challenge: The Data-Driven Radio Frequency Signal Separation Challenge”
Correction to “RF Challenge: The Data-Driven Radio Frequency Signal Separation Challenge” Open
Presents corrections to the paper, (Correction to “RF Challenge: The Data-Driven Radio Frequency Signal Separation Challenge”).
View article: RF Challenge: The Data-Driven Radio Frequency Signal Separation Challenge
RF Challenge: The Data-Driven Radio Frequency Signal Separation Challenge Open
We address the critical problem of interference rejection in radio-frequency (RF) signals using a data-driven approach that leverages deep-learning methods. A primary contribution of this paper is the introduction of the RF Challenge, whic…
View article: RF Challenge: The Data-Driven Radio Frequency Signal Separation Challenge
RF Challenge: The Data-Driven Radio Frequency Signal Separation Challenge Open
We address the critical problem of interference rejection in radio-frequency (RF) signals using a data-driven approach that leverages deep-learning methods. A primary contribution of this paper is the introduction of the RF Challenge, whic…
View article: Score-based Source Separation with Applications to Digital Communication Signals
Score-based Source Separation with Applications to Digital Communication Signals Open
We propose a new method for separating superimposed sources using diffusion-based generative models. Our method relies only on separately trained statistical priors of independent sources to establish a new objective function guided by max…
View article: Self-Supervised Representations for Singing Voice Conversion
Self-Supervised Representations for Singing Voice Conversion Open
A singing voice conversion model converts a song in the voice of an arbitrary source singer to the voice of a target singer. Recently, methods that leverage self-supervised audio representations such as HuBERT and Wav2Vec 2.0 have helped f…
View article: Detecting Audio Attacks on ASR Systems with Dropout Uncertainty
Detecting Audio Attacks on ASR Systems with Dropout Uncertainty Open
Various adversarial audio attacks have recently been developed to fool automatic speech recognition (ASR) systems. We here propose a defense against such attacks based on the uncertainty introduced by dropout in neural networks. We show th…
View article: LAP-based motion-compensated frame interpolation
LAP-based motion-compensated frame interpolation Open
High-quality video frame interpolation often necessitates accurate motion estimates between consecutive frames. Standard video encoding schemes often
\nestimate the motion between frames using variants of block matching algorithms. For the…