A5017449169- Speech and Audio Processing
- Advanced Adaptive Filtering Techniques
- Complex Network Analysis Techniques
- Music and Audio Processing
- Hearing Loss and Rehabilitation
- Speech Recognition and Synthesis
- Advanced Graph Neural Networks
- Indoor and Outdoor Localization Technologies
- Graph Theory and Algorithms
- Reinforcement Learning in Robotics
- Bayesian Modeling and Causal Inference
- Opinion Dynamics and Social Influence
- Acoustic Wave Phenomena Research
- Data Mining Algorithms and Applications
- Advanced Machining and Optimization Techniques
- Blind Source Separation Techniques
- Artificial Intelligence in Games
- Advanced Data Compression Techniques
- Data Management and Algorithms
- Advanced machining processes and optimization
- Sports Analytics and Performance
- Analytical Chemistry and Chromatography
- Bioinformatics and Genomic Networks
- Ergonomics and Musculoskeletal Disorders
- Human-Automation Interaction and Safety
Bowie State University
2023-2025
University of Arizona
2023-2024
University of Oregon
2023
University of North Texas
2023
Virginia Commonwealth University
2021-2022
Oregon State University
2022
University of California, Santa Cruz
2019-2021
Microsoft (United States)
2010-2021
Microsoft (Finland)
2020-2021
Intel (United States)
1988-2021
The INTERSPEECH 2020 Deep Noise Suppression (DNS) Challenge is intended to promote collaborative research in realtime single-channel Speech Enhancement aimed maximize the subjective (perceptual) quality of enhanced speech.A typical approach evaluate noise suppression methods use objective metrics on test set obtained by splitting original dataset.While performance good synthetic set, often model degrades significantly real recordings.Also, most conventional do not correlate well with tests...
The Deep Noise Suppression (DNS) challenge was designed to unify the research efforts in area of noise suppression targeted for human perception.We recently organized a DNS special session at INTERSPEECH 2020 and ICASSP 2021.We open-sourced training test datasets wideband scenario along with subjective evaluation framework based on ITU-T standard P.808, which used evaluate participants challenge.Many researchers from academia industry made significant contributions push field forward, yet...
In this paper, we present a new technique for the estimation of short-term linear predictive parameters speech and noise from noisy data their subsequent use in waveform enhancement schemes. The method exploits priori information about spectral shapes stored trained codebooks, parameterized as coefficients. also uses statistics estimated observation. Maximum-likelihood estimates predictor are obtained by searching combination codebook entries that optimizes likelihood. involves computation...
We consider an agent's uncertainty about its environment and the problem of generalizing this across observations. Specifically, we focus on exploration in non-tabular reinforcement learning. Drawing inspiration from intrinsic motivation literature, use density models to measure uncertainty, propose a novel algorithm for deriving pseudo-count arbitrary model. This technique enables us generalize count-based algorithms case. apply our ideas Atari 2600 games, providing sensible pseudo-counts...
Despite the prevalence of community detection algorithms, relatively less work has been done on understanding whether a network is indeed modular and how resilient structure under perturbations. To address this issue, we propose new vertex-based metric called "permanence", that can quantitatively give an estimate community- like network.
Background noise is a major source of quality impairments in Voice over Internet Protocol (VoIP) and Public Switched Telephone Network (PSTN) calls.Recent work shows the efficacy deep learning for suppression, but datasets have been relatively small compared to those used other domains (e.g., ImageNet) associated evaluations more focused.In order better facilitate research Speech Enhancement, we present noisy speech dataset (MS-SNSD) that can scale arbitrary sizes depending on number...
The Deep Noise Suppression (DNS) challenge is designed to foster innovation in the area of noise suppression achieve superior perceptual speech quality. We recently organized a DNS special session at INTERSPEECH 2020 where we open-sourced training and test datasets for researchers train their models. also subjective evaluation framework used tool evaluate select final winners. Many from academia industry made significant contributions push field forward. learned that as research community,...
OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning search/planning games. supports n-player (single- multi- agent) zero-sum, cooperative general-sum, one-shot sequential, strictly turn-taking simultaneous-move, perfect imperfect information games, as well traditional multiagent such (partially- fully- observable) grid worlds social dilemmas. also includes tools to analyze dynamics other common evaluation metrics. This document serves both...
Speech enhancement is challenging because of the diversity background noise types. Most existing methods are focused on modelling speech rather than noise. In this paper, we propose a novel idea to model and simultaneously in two-branch convolutional neural network, namely SN-Net. SN-Net, two branches predict noise, respectively. Instead information fusion only at final output layer, interaction modules introduced several intermediate feature domains between benefit each other. Such an can...
The INTERSPEECH 2020 Deep Noise Suppression Challenge is intended to promote collaborative research in real-time single-channel Speech Enhancement aimed maximize the subjective (perceptual) quality of enhanced speech. A typical approach evaluate noise suppression methods use objective metrics on test set obtained by splitting original dataset. Many publications report reasonable performance synthetic drawn from same distribution as that training set. However, often model degrades...
Significant growth in Electronic Health Records (EHR) over the last decade has provided an abundance of clinical text that is mostly unstructured and untapped. This huge amount data motivated development new information extraction mining techniques. Named Entity Recognition (NER) Relationship Extraction (RE) are key components tasks domain. In this paper, we highlight present status NER RE techniques detail by discussing existing proposed NLP models for two their performances discuss current...
CyVerse, the largest publicly-funded open-source research cyberinfrastructure for life sciences, has played a crucial role in advancing data-driven since 2010s. As technology landscape evolved with emergence of cloud computing platforms, machine learning and artificial intelligence (AI) applications, CyVerse enabled access by providing interfaces, Software as Service (SaaS), cloud-native Infrastructure Code (IaC) to leverage new technologies. services enable researchers integrate...
In this paper, we propose a Bayesian minimum mean squared error approach for the joint estimation of short-term predictor parameters speech and noise, from noisy observation. We use trained codebooks noise linear predictive coefficients to model priori information required by scheme. contrast current approaches that consider excitation variances as part information, in proposed method they are computed online each short-time segment, based on observation at hand. Consequently, performs well...
This article provides an overview of the embedded multidie interconnect bridge (EMIB) multichip packaging (MCP) technology. EMIB is a unique paradigm that very high-density interconnects (currently in range 500-1000 I/O/mm) localized between two devices, thus enabling high-bandwidth (BW) on-package links while leaving rest package structures and designs unaffected. The construction silicon to allow high BW electrical signaling dies discussed detail. Examples implementations for...
Optimization of parameterized policies for reinforcement learning (RL) is an important and challenging problem in artificial intelligence. Among the most common approaches are algorithms based on gradient ascent a score function representing discounted return. In this paper, we examine role these policy actor-critic partially-observable multiagent environments. We show several candidate update rules relate them to foundation regret minimization techniques one-shot tabular cases, leading...
Identifying community structure is a fundamental problem in network analysis. Most detection algorithms are based on optimizing combinatorial parameter, for example modularity. This optimization generally NP-hard, thus merely changing the vertex order can alter their assignments to community. However, there has been less study how ordering influences results of algorithms. Here we identify and properties invariant groups vertices (constant communities) whose assignment communities are, quite...
The package integrated inductors employed by Intel's Fully Integrated Voltage Regulator (FIVR) have had to scale in tandem with the circuits they power. This reduction available volume has resulted degraded inductor quality factor and reduced conversion efficiency. In this work, a novel coaxial magnetic composite core inductor, called Coax MIL, is described. MIL dramatically improves performance factor. improvement metrics translates ripple current enhanced