Bandwidth estimation and congestion control for real-time communications (i.e., audio video conferencing) remains a difficult problem, despite many years of research. Achieving high quality experience (QoE) end users requires continual updates due to changing network architectures technologies. In this paper, we apply reinforcement learning the first time problem (RTC), where seek optimize user-perceived quality. We present initial proof-of-concept results, learn an agent sending rate in RTC...

Commonly used datasets for evaluating video codecs are all very high quality and not representative of typically in conferencing scenarios. We present the Video Conferencing Dataset (VCD) real-time communication, first such dataset focused on conferencing. VCD includes a wide variety camera qualities spatial temporal information. It both desktop mobile scenarios two types background processing. report compression efficiency H.264, H.265, H.266, AV1 low-delay settings compare it with...

Workplace meetings are vital to organizational collaboration, yet relatively little progress has been made toward measuring meeting effectiveness and inclusiveness at scale. The recent rise in remote hybrid represents an opportunity do so via computer-mediated communication (CMC) systems. Here, we share the results of effective inclusive survey embedded within a CMC system diverse set companies organizations. We correlate with objective metrics available from identify generalizable...

Classic public switched telephone networks (PSTN) are often a black box for VoIP network providers, as they have no access to performance indicators, such delay or packet loss.Only the degraded output speech signal can be used monitor quality of these networks.However, current state-of-the-art models not reliable enough live monitoring.One reasons this is that PSTN distortions unique depending on provider and country, which makes it difficult train model generalizes well different...

User-perceived quality-of-experience (QoE) in internet telephony systems is commonly evaluated using subjective ratings computed as a Mean Opinion Score (MOS). In such systems, while user MOS can be tracked on an ongoing basis, it does not give insight into which factors of call induced any perceived degradation QoE - tell us what caused to have sub-optimal experience. For effective planning product improvements, we are interested understanding the impact each these degrading factors,...

Failure to accurately measure the outcomes of an experiment can lead bias and incorrect conclusions. Online controlled experiments (aka AB tests) are increasingly being used make decisions improve websites as well mobile desktop applications. We argue that loss telemetry data (during upload or post-processing) skew results experiments, leading statistical power inaccurate erroneous By systematically investigating causes loss, we it is not practical entirely eliminate it. Consequently,...

Meetings are a pervasive method of communication within all types companies and organizations, using remote collaboration systems to conduct meetings has increased dramatically since the COVID-19 pandemic. However, not inclusive, especially in terms participation rates among attendees. In recent large-scale survey conducted at Microsoft, top suggestion given by meeting participants for improving inclusiveness is improve ability interrupt acquire floor during meetings. We show that use...

Conventional search engines such as Bing and Google provide a user with short answer to some queries well ranked list of documents, in order better meet her information needs. In this paper we study class that call math. Calculations (e.g. "12% 24$ ", "square root 120"), unit conversions "convert 10 meter feet"), symbolic computations "plot x^2+x+1") are examples math queries. Among the should be answered, special because infinite combinations numbers symbols, rather few keywords form them....

User experience in real-time video applications requires continuously adjusting encoding bitrates to match available network capacity, which hinges on accurate bandwidth estimation (BWE). However, heterogeneity prevents a one-size-fits-all solution BWE, motivating the demand for personalized approaches. Although personalizing BWE algorithms offers benefits such as improved adaptability individual conditions, it faces challenge of data drift -- where estimators degrade over time due evolving...

The difficulty of exploring and training online on real production systems limits the scope real-time data/feedback-driven decision making. most feasible approach is to adopt offline reinforcement learning from limited trajectory samples. However, after deployment, such policies fail due exogenous factors that temporarily or permanently disturb/alter transition distribution assumed process structure induced by This results in critical policy failures generalization errors sensitive domains...

Machine learning-based video codecs have made significant progress in the past few years. A critical area development of ML-based is an accurate evaluation metric that does not require expensive and slow subjective test. We show existing metrics were designed trained on DSP-based are highly correlated to opinion when used with ML due artifacts being quite different between codecs. provide a new dataset codec videos been accurately labeled for quality. also propose full reference quality...

Abstract A stochastic graph process with a Markov property is introduced to model the flow of an infectious disease over known contact network. The provides probability distribution unobserved pathways. basic reproductive number in compartmental models generalized dynamic based on sequence outdegrees process. cumulative resistance and threat associated each individual also measured indegree outdegree applied outbreak data from 2001 foot‐and‐mouth (FMD) United Kingdom. Canadian Journal...

A primary goal of remote collaboration tools is to provide effective and inclusive meetings for all participants. To study meeting effectiveness inclusiveness, we first conducted a large-scale email survey (N=4,425; after filtering N=3,290) at large technology company (pre-COVID-19); using this data derived multivariate model show how it correlates with participation, feeling comfortable contribute. We believe the such inclusiveness. The size provided opportunity analyze correlations that...

Principal components analysis is a well-known statistical method in dealing with large dependent data sets. It also used functional for both purposes of reduction as well variation representation. On the other hand "handwriting" one objects, studied various fields like pattern recognition and shape analysis. Considering time argument, handwriting would be an infinite dimensional data; object. In this paper we try to use principal (FPCA) Persian data, analyzing word Mehr which term Love.

User surveys for Quality of Experience (QoE) are a critical source information application developers. In addition to the common "star rating" used estimate Mean Opinion Score (MOS), more detailed survey questions (problem tokens) about specific areas provide valuable insight into factors impacting QoE. This paper explores two aspects problem token questionnaire design. First, we study bias introduced by fixed question order, and second, methodology manage size while keeping it informative....

Meetings are a pervasive method of communication within all types companies and organizations, using remote collaboration systems to conduct meetings has increased dramatically since the COVID-19 pandemic. However, not inclusive, especially in terms participation rates among attendees. In recent large-scale survey conducted at Microsoft, top suggestion given by meeting participants for improving inclusiveness is improve ability interrupt acquire floor during meetings. We show that use...

Workplace meetings are vital to organizational collaboration, yet relatively little progress has been made toward measuring meeting effectiveness and inclusiveness at scale. The recent rise in remote hybrid represents an opportunity do so via computer-mediated communication (CMC) systems. Here, we share the results of effective inclusive survey embedded within a CMC system diverse set companies organizations. We correlate with objective metrics available from identify generalizable...

Commonly used datasets for evaluating video codecs are all very high quality and not representative of typically in conferencing scenarios. We present the Video Conferencing Dataset (VCD) real-time communication, first such dataset focused on conferencing. VCD includes a wide variety camera qualities spatial temporal information. It both desktop mobile scenarios two types background processing. report compression efficiency H.264, H.265, H.266, AV1 low-delay settings compare it with...

Classic public switched telephone networks (PSTN) are often a black box for VoIP network providers, as they have no access to performance indicators, such delay or packet loss. Only the degraded output speech signal can be used monitor quality of these networks. However, current state-of-the-art models not reliable enough live monitoring. One reasons this is that PSTN distortions unique depending on provider and country, which makes it difficult train model generalizes well different In...