arXiv subjects

Arpit Gupta

Publications and source records attributed to Arpit Gupta.

At least 37 records · Page 2Linked to original sources

arXiv2024

The Efficacy of the Connect America Fund in Addressing US Internet Access Inequities

Residential fixed broadband internet access in the United States (US) has long been distributed inequitably, drawing significant attention from researchers and policymakers. This paper evaluates the efficacy of the Connect America Fund (CAF), a key policy intervention aimed at addressing disparities in US internet access. CAF subsidizes the creation of new regulated broadband monopolies in underserved areas, aiming to provide comparable internet access, in terms of price and speed, to that available in urban regions. Oversight of CAF largely relies on data self-reported by internet service providers (ISPs), which is often questionable. We use the broadband-plan querying tool (BQT) to curate a novel dataset that complements ISP-reported information with ISP-advertised broadband plan details (download speed and monthly cost) on publicly accessible websites. Specifically, we query advertised broadband plans for 687k residential addresses across 15 states, certified as served by ISPs to regulators. Our analysis reveals significant discrepancies between ISP-reported data and actual broadband availability. We find that the serviceability rate-defined as the fraction of addresses ISPs actively serve out of the total queried, weighted by the number of CAF addresses in a census block group-is only 55%, dropping to as low as 18% in some states. Additionally, the compliance rate-defined as the weighted fraction of addresses where ISPs actively serve and advertise download speeds above the FCC's 10 Mbps threshold-is only 33%. We also observe that in a subset of census blocks, CAF-funded addresses receive higher broadband speeds than their monopoly-served neighbors. These results indicate that while a few users have benefited from this multi-billion dollar program, it has largely failed to achieve its intended goal, leaving many targeted rural communities with inadequate or no broadband connectivity.

cs.NI↗

arXiv2024

Mitigating Bias for Question Answering Models by Tracking Bias Influence

Models of various NLP tasks have been shown to exhibit stereotypes, and the bias in the question answering (QA) models is especially harmful as the output answers might be directly consumed by the end users. There have been datasets to evaluate bias in QA models, while bias mitigation technique for the QA models is still under-explored. In this work, we propose BMBI, an approach to mitigate the bias of multiple-choice QA models. Based on the intuition that a model would lean to be more biased if it learns from a biased example, we measure the bias level of a query instance by observing its influence on another instance. If the influenced instance is more biased, we derive that the query instance is biased. We then use the bias level detected as an optimization objective to form a multi-task learning setting in addition to the original QA task. We further introduce a new bias evaluation metric to quantify bias in a comprehensive and sensitive way. We show that our method could be applied to multiple QA formulations across multiple bias categories. It can significantly reduce the bias level in all 9 bias categories in the BBQ dataset while maintaining comparable QA accuracy.

cs.CL↗

arXiv2024

DiNADO: Norm-Disentangled Neurally-Decomposed Oracles for Controlling Language Models

NeurAlly-Decomposed Oracle (NADO) is a powerful approach for controllable generation with large language models. It is designed to avoid catastrophic forgetting while achieving guaranteed convergence to an entropy-maximized closed-form optimal solution with reasonable modeling capacity. Despite the success, several challenges arise when apply NADO to a wide range of scenarios. Vanilla NADO suffers from gradient vanishing for low-probability control signals and is highly reliant on a regularization to satisfy the stochastic version of Bellman equation. In addition, the vanilla implementation of NADO introduces a few additional transformer layers, suffering from a limited capacity especially compared to other finetune-based model adaptation methods like LoRA. In this paper, we propose a improved version of the NADO algorithm, namely DiNADO (norm-Disentangled NeurAlly-Decomposed Oracles), which improves the performance of the NADO algorithm through disentangling the step-wise global norm over the approximated oracle $R$-value for all potential next-tokens, allowing DiNADO to be combined with finetuning methods like LoRA. We discuss in depth how DiNADO achieves better capacity, stability and flexibility with both empirical and theoretical results. Experiments on formality control in machine translation and the lexically constrained generation task CommonGen demonstrates the significance of the improvements.

cs.CL↗

arXiv2023

Analyzing Disparity and Temporal Progression of Internet Quality through Crowdsourced Measurements with Bias-Correction

Crowdsourced speedtest measurements are an important tool for studying internet performance from the end user perspective. Nevertheless, despite the accuracy of individual measurements, simplistic aggregation of these data points is problematic due to their intrinsic sampling bias. In this work, we utilize a dataset of nearly 1 million individual Ookla Speedtest measurements, correlate each datapoint with 2019 Census demographic data, and develop new methods to present a novel analysis to quantify regional sampling bias and the relationship of internet performance to demographic profile. We find that the crowdsourced Ookla Speedtest data points contain significant sampling bias across different census block groups based on a statistical test of homogeneity. We introduce two methods to correct the regional bias by the population of each census block group. Whereas the sampling bias leads to a small discrepancy in the overall cumulative distribution function of internet speed in a city between estimation from original samples and bias-corrected estimation, the discrepancy is much smaller compared to the size of the sampling heterogeneity across regions. Further, we show that the sampling bias is strongly associated with a few demographic variables, such as income, education level, age, and ethnic distribution. Through regression analysis, we find that regions with higher income, younger populations, and lower representation of Hispanic residents tend to measure faster internet speeds along with substantial collinearity amongst socioeconomic attributes and ethnic composition. Finally, we find that average internet speed increases over time based on both linear and nonlinear analysis from state space models, though the regional sampling bias may result in a small overestimation of the temporal increase of internet speed.

stat.AP↗

arXiv2023

Estimating WebRTC Video QoE Metrics Without Using Application Headers

The increased use of video conferencing applications (VCAs) has made it critical to understand and support end-user quality of experience (QoE) by all stakeholders in the VCA ecosystem, especially network operators, who typically do not have direct access to client software. Existing VCA QoE estimation methods use passive measurements of application-level Real-time Transport Protocol (RTP) headers. However, a network operator does not always have access to RTP headers in all cases, particularly when VCAs use custom RTP protocols (e.g., Zoom) or due to system constraints (e.g., legacy measurement systems). Given this challenge, this paper considers the use of more standard features in the network traffic, namely, IP and UDP headers, to provide per-second estimates of key VCA QoE metrics such as frames rate and video resolution. We develop a method that uses machine learning with a combination of flow statistics (e.g., throughput) and features derived based on the mechanisms used by the VCAs to fragment video frames into packets. We evaluate our method for three prevalent VCAs running over WebRTC: Google Meet, Microsoft Teams, and Cisco Webex. Our evaluation consists of 54,696 seconds of VCA data collected from both (1), controlled in-lab network conditions, and (2) real-world networks from 15 households. We show that the ML-based approach yields similar accuracy compared to the RTP-based methods, despite using only IP/UDP data. For instance, we can estimate FPS within 2 FPS for up to 83.05% of one-second intervals in the real-world data, which is only 1.76% lower than using the application-level RTP headers.

cs.NI↗

arXiv2023

In Search of netUnicorn: A Data-Collection Platform to Develop Generalizable ML Models for Network Security Problems

The remarkable success of the use of machine learning-based solutions for network security problems has been impeded by the developed ML models' inability to maintain efficacy when used in different network environments exhibiting different network behaviors. This issue is commonly referred to as the generalizability problem of ML models. The community has recognized the critical role that training datasets play in this context and has developed various techniques to improve dataset curation to overcome this problem. Unfortunately, these methods are generally ill-suited or even counterproductive in the network security domain, where they often result in unrealistic or poor-quality datasets. To address this issue, we propose an augmented ML pipeline that leverages explainable ML tools to guide the network data collection in an iterative fashion. To ensure the data's realism and quality, we require that the new datasets should be endogenously collected in this iterative process, thus advocating for a gradual removal of data-related problems to improve model generalizability. To realize this capability, we develop a data-collection platform, netUnicorn, that takes inspiration from the classic "hourglass" model and is implemented as its "thin waist" to simplify data collection for different learning problems from diverse network environments. The proposed system decouples data-collection intents from the deployment mechanisms and disaggregates these high-level intents into smaller reusable, self-contained tasks. We demonstrate how netUnicorn simplifies collecting data for different learning problems from multiple network environments and how the proposed iterative data collection improves a model's generalizability.

cs.NI↗

arXiv2023

Watching Stars in Pixels: The Interplay of Traffic Shaping and YouTube Streaming QoE over GEO Satellite Networks

Geosynchronous satellite (GEO) networks are a crucial option for users beyond terrestrial connectivity. However, unlike terrestrial networks, GEO networks exhibit high latency and deploy TCP proxies and traffic shapers. The deployment of proxies effectively mitigates the impact of high network latency in GEO networks, while traffic shapers help realize customer-controlled data-saver options that optimize data usage. It is unclear how the interplay between GEO networks' high latency, TCP proxies, and traffic-shaping policies affects the quality of experience (QoE) for commonly used video applications. To fill this gap, we analyze the quality of over $2$k YouTube video sessions streamed across a production GEO network with a $900$Kbps shaping rate. Given the average bit rates for the selected videos, we expected seamless streaming at $360$p or lower resolutions. However, our analysis reveals that this is not the case: $28\%$ of TCP sessions and $18\%$ of gQUIC sessions experience rebuffering events, while the median average resolution is only $380$p for TCP and $299$p for gQUIC. Our analysis identifies two key factors contributing to sub-optimal performance: (i)unlike TCP, gQUIC only utilizes $63\%$ of network capacity; and (ii)YouTube's imperfect chunk request pipelining. As a result of our study, the partner GEO ISP discontinued support for the low-bandwidth data-saving option in U.S. business and residential markets to avoid potential degradation of video quality -- highlighting the practical significance of our findings.

cs.NI↗

arXiv2023

Parameter-Efficient Low-Resource Dialogue State Tracking by Prompt Tuning

Dialogue state tracking (DST) is an important step in dialogue management to keep track of users' beliefs. Existing works fine-tune all language model (LM) parameters to tackle the DST task, which requires significant data and computing resources for training and hosting. The cost grows exponentially in the real-world deployment where dozens of fine-tuned LM are used for different domains and tasks. To reduce parameter size and better utilize cross-task shared information, we propose to use soft prompt token embeddings to learn task properties. Without tuning LM parameters, our method drastically reduces the number of parameters needed to less than 0.5% of prior works while achieves better low-resource DST performance.

cs.CL↗

arXiv2023

Challenges and Opportunities for Beyond-5G Wireless Security

The demand for broadband wireless access is driving research and standardization of 5G and beyond-5G wireless systems. In this paper, we aim to identify emerging security challenges for these wireless systems and pose multiple research areas to address these challenges.

eess.SY↗

arXiv2023

Decoding the Divide: Analyzing Disparities in Broadband Plans Offered by Major US ISPs

Digital equity in Internet access is often measured along three axes: availability, affordability, and adoption. Most prior work focuses on availability; the other two aspects have received little attention. In this paper, we study broadband affordability in the US. Specifically, we focus on the nature of broadband plans offered by major ISPs across the US. To this end, we develop a broadband plan querying tool (BQT) that obtains broadband plans (upload/download speed and price) offered by seven major ISPs for any street address in the US. We then use this tool to curate a dataset, querying broadband plans for over 837k street addresses in thirty cities for seven major wireline broadband ISPs. We use a plan's carriage value, the Mbps of a user's traffic that an ISP carries for one dollar, to compare plans. Our analysis provides us with the following new insights: (1) ISP plans vary inter-city. Specifically, the fraction of census block groups that receive high and low carriage value plans varies widely by city; (2) ISP plans intra-city are spatially clustered, and the carriage value can vary as much as 600% within a city; (3) Cable-based ISPs offer up to 30% more carriage value to users when competing with fiber-based ISPs in a block group; and (4) Average income in a block group plays a critical role in dictating who gets a fiber deployment (i.e., a better carriage value) in the US. While we hope our tool, dataset, and analysis in their current form are helpful for policymakers at different levels (city, county, state), they are only a small step toward understanding digital equity. Based on our learnings, we conclude with recommendations to continue to advance our understanding of broadband affordability.

cs.NI↗

arXiv2021

Characterizing Performance Inequity Across U.S. Ookla Speedtest Users

The Internet has become indispensable to daily activities, such as work, education and health care. Many of these activities require Internet access data rates that support real-time video conferencing. However, digital inequality persists across the United States, not only in who has access but in the quality of that access. Speedtest by Ookla allows users to run network diagnostic tests to better understand the current performance of their network. In this work, we leverage an Internet performance dataset from Ookla, together with an ESRI demographic dataset, to conduct a comprehensive analysis that characterizes performance differences between Speedtest users across the U.S. Our analysis shows that median download speeds for Speedtest users can differ by over 150Mbps between states. Further, there are important distinctions between user categories. For instance, all but one state showed statistically significant differences in performance between Speedtest users in urban and rural areas. The difference also exists in urban areas between high and low income users in 27 states. Our analysis reveals that states that demonstrate this disparity in Speedtest results are geographically bigger, more populous and have a wider dispersion of median household income. We conclude by highlighting several challenges to the complex problem space of digital inequality characterization and provide recommendations for furthering research on this topic.

cs.NI↗

arXiv2021

DynamiQ: Planning for Dynamics in Network Streaming Analytics Systems

The emergence of programmable data-plane targets has motivated a new hybrid design for network streaming analytics systems that combine these targets' fast packet processing speeds with the rich compute resources available at modern stream processors. However, these systems require careful query planning; that is, specifying the minute details of executing a given set of queries in a way that makes the best use of the limited resources and programmability offered by data-plane targets. We use such an existing system, Sonata, and real-world packet traces to understand how executing a fixed query workload is affected by the unknown dynamics of the traffic that defines the target's input workload. We observe that static query planning, as employed by Sonata, cannot handle even small changes in the input workload, wasting data-plane resources to the point where query execution is confined mainly to userspace. This paper presents the design and implementation of DynamiQ, a new network streaming analytics platform that employs dynamic query planning to deal with the dynamics of real-world input workloads. Specifically, we develop a suite of practical algorithms for (i) computing effective initial query plans (to start query execution) and (ii) enabling efficient updating of portions of such an initial query plan at runtime (to adapt to changes in the input workload). Using real-world packet traces as input workload, we show that compared to Sonata, DynamiQ reduces the stream processor's workload by two orders of magnitude.

cs.NI↗

arXiv2020

Exposure Density and Neighborhood Disparities in COVID-19 Infection Risk: Using Large-scale Geolocation Data to Understand Burdens on Vulnerable Communities

This study develops a new method to quantify neighborhood activity levels at high spatial and temporal resolutions and test whether, and to what extent, behavioral responses to social distancing policies vary with socioeconomic and demographic characteristics. We define exposure density as a measure of both the localized volume of activity in a defined area and the proportion of activity occurring in non-residential and outdoor land uses. We utilize this approach to capture inflows/outflows of people as a result of the pandemic and changes in mobility behavior for those that remain. First, we develop a generalizable method for assessing neighborhood activity levels by land use type using smartphone geolocation data over a three-month period covering more than 12 million unique users within the Greater New York area. Second, we measure and analyze disparities in community social distancing by identifying patterns in neighborhood activity levels and characteristics before and after the stay-at-home order. Finally, we evaluate the effect of social distancing in neighborhoods on COVID-19 infection rates and outcomes associated with localized demographic, socioeconomic, and infrastructure characteristics in order to identify disparities in health outcomes related to exposure risk. Our findings provide insight into the timely evaluation of the effectiveness of social distancing for individual neighborhoods and support a more equitable allocation of resources to support vulnerable and at-risk communities. Our findings demonstrate distinct patterns of activity pre- and post-COVID across neighborhoods. The variation in exposure density has a direct and measurable impact on the risk of infection.

cs.CY↗

arXiv2019

A dataset for resolving referring expressions in spoken dialogue via contextual query rewrites (CQR)

We present Contextual Query Rewrite (CQR) a dataset for multi-domain task-oriented spoken dialogue systems that is an extension of the Stanford dialog corpus (Eric et al., 2017a). While previous approaches have addressed the issue of diverse schemas by learning candidate transformations (Naik et al., 2018), we instead model the reference resolution task as a user query reformulation task, where the dialog state is serialized into a natural language query that can be executed by the downstream spoken language understanding system. In this paper, we describe our methodology for creating the query reformulation extension to the dialog corpus, and present an initial set of experiments to establish a baseline for the CQR task. We have released the corpus to the public [1] to support further research in this area.

cs.CL↗

arXiv2019

Scaling Multi-Domain Dialogue State Tracking via Query Reformulation

We present a novel approach to dialogue state tracking and referring expression resolution tasks. Successful contextual understanding of multi-turn spoken dialogues requires resolving referring expressions across turns and tracking the entities relevant to the conversation across turns. Tracking conversational state is particularly challenging in a multi-domain scenario when there exist multiple spoken language understanding (SLU) sub-systems, and each SLU sub-system operates on its domain-specific meaning representation. While previous approaches have addressed the disparate schema issue by learning candidate transformations of the meaning representation, in this paper, we instead model the reference resolution as a dialogue context-aware user query reformulation task -- the dialog state is serialized to a sequence of natural language tokens representing the conversation. We develop our model for query reformulation using a pointer-generator network and a novel multi-task learning setup. In our experiments, we show a significant improvement in absolute F1 on an internal as well as a, soon to be released, public benchmark respectively.

cs.CL↗

arXiv2018

Cross-Lingual Approaches to Reference Resolution in Dialogue Systems

In the slot-filling paradigm, where a user can refer back to slots in the context during the conversation, the goal of the contextual understanding system is to resolve the referring expressions to the appropriate slots in the context. In this paper, we build on the context carryover system~\citep{Naik2018ContextualSC}, which provides a scalable multi-domain framework for resolving references. However, scaling this approach across languages is not a trivial task, due to the large demand on acquisition of annotated data in the target language. Our main focus is on cross-lingual methods for reference resolution as a way to alleviate the need for annotated data in the target language. In the cross-lingual setup, we assume there is access to annotated resources as well as a well trained model in the source language and little to no annotated data in the target language. In this paper, we explore three different approaches for cross-lingual transfer \textemdash~\ delexicalization as data augmentation, multilingual embeddings and machine translation. We compare these approaches both on a low resource setting as well as a large resource setting. Our experiments show that multilingual embeddings and delexicalization via data augmentation have a significant impact in the low resource setting, but the gains diminish as the amount of available data in the target language increases. Furthermore, when combined with machine translation we can get performance very close to actual live data in the target language, with only 25\% of the data projected into the target language.

cs.CL↗

arXiv2018

Contextual Slot Carryover for Disparate Schemas

In the slot-filling paradigm, where a user can refer back to slots in the context during a conversation, the goal of the contextual understanding system is to resolve the referring expressions to the appropriate slots in the context. In large-scale multi-domain systems, this presents two challenges - scaling to a very large and potentially unbounded set of slot values, and dealing with diverse schemas. We present a neural network architecture that addresses the slot value scalability challenge by reformulating the contextual interpretation as a decision to carryover a slot from a set of possible candidates. To deal with heterogenous schemas, we introduce a simple data-driven method for trans- forming the candidate slots. Our experiments show that our approach can scale to multiple domains and provides competitive results over a strong baseline.

cs.CL↗

arXiv2018

Just ASK: Building an Architecture for Extensible Self-Service Spoken Language Understanding

This paper presents the design of the machine learning architecture that underlies the Alexa Skills Kit (ASK) a large scale Spoken Language Understanding (SLU) Software Development Kit (SDK) that enables developers to extend the capabilities of Amazon's virtual assistant, Alexa. At Amazon, the infrastructure powers over 25,000 skills deployed through the ASK, as well as AWS's Amazon Lex SLU Service. The ASK emphasizes flexibility, predictability and a rapid iteration cycle for third party developers. It imposes inductive biases that allow it to learn robust SLU models from extremely small and sparse datasets and, in doing so, removes significant barriers to entry for software developers and dialogue systems researchers.

cs.CL↗