-
Neural Machine Translation for the Indigenous Languages of the Americas: An Introduction
Authors:
Manuel Mager,
Rajat Bhatnagar,
Graham Neubig,
Ngoc Thang Vu,
Katharina Kann
Abstract:
Neural models have drastically advanced state of the art for machine translation (MT) between high-resource languages. Traditionally, these models rely on large amounts of training data, but many language pairs lack these resources. However, an important part of the languages in the world do not have this amount of data. Most languages from the Americas are among them, having a limited amount of p…
▽ More
Neural models have drastically advanced state of the art for machine translation (MT) between high-resource languages. Traditionally, these models rely on large amounts of training data, but many language pairs lack these resources. However, an important part of the languages in the world do not have this amount of data. Most languages from the Americas are among them, having a limited amount of parallel and monolingual data, if any. Here, we present an introduction to the interested reader to the basic challenges, concepts, and techniques that involve the creation of MT systems for these languages. Finally, we discuss the recent advances and findings and open questions, product of an increased interest of the NLP community in these languages.
△ Less
Submitted 11 June, 2023;
originally announced June 2023.
-
Human-AI Collaboration via Conditional Delegation: A Case Study of Content Moderation
Authors:
Vivian Lai,
Samuel Carton,
Rajat Bhatnagar,
Q. Vera Liao,
Yunfeng Zhang,
Chenhao Tan
Abstract:
Despite impressive performance in many benchmark datasets, AI models can still make mistakes, especially among out-of-distribution examples. It remains an open question how such imperfect models can be used effectively in collaboration with humans. Prior work has focused on AI assistance that helps people make individual high-stakes decisions, which is not scalable for a large amount of relatively…
▽ More
Despite impressive performance in many benchmark datasets, AI models can still make mistakes, especially among out-of-distribution examples. It remains an open question how such imperfect models can be used effectively in collaboration with humans. Prior work has focused on AI assistance that helps people make individual high-stakes decisions, which is not scalable for a large amount of relatively low-stakes decisions, e.g., moderating social media comments. Instead, we propose conditional delegation as an alternative paradigm for human-AI collaboration where humans create rules to indicate trustworthy regions of a model. Using content moderation as a testbed, we develop novel interfaces to assist humans in creating conditional delegation rules and conduct a randomized experiment with two datasets to simulate in-distribution and out-of-distribution scenarios. Our study demonstrates the promise of conditional delegation in improving model performance and provides insights into design for this novel paradigm, including the effect of AI explanations.
△ Less
Submitted 25 April, 2022;
originally announced April 2022.
-
Don't Rule Out Monolingual Speakers: A Method For Crowdsourcing Machine Translation Data
Authors:
Rajat Bhatnagar,
Ananya Ganesh,
Katharina Kann
Abstract:
High-performing machine translation (MT) systems can help overcome language barriers while making it possible for everyone to communicate and use language technologies in the language of their choice. However, such systems require large amounts of parallel sentences for training, and translators can be difficult to find and expensive. Here, we present a data collection strategy for MT which, in co…
▽ More
High-performing machine translation (MT) systems can help overcome language barriers while making it possible for everyone to communicate and use language technologies in the language of their choice. However, such systems require large amounts of parallel sentences for training, and translators can be difficult to find and expensive. Here, we present a data collection strategy for MT which, in contrast, is cheap and simple, as it does not require bilingual speakers. Based on the insight that humans pay specific attention to movements, we use graphics interchange formats (GIFs) as a pivot to collect parallel sentences from monolingual annotators. We use our strategy to collect data in Hindi, Tamil and English. As a baseline, we also collect data using images as a pivot. We perform an intrinsic evaluation by manually evaluating a subset of the sentence pairs and an extrinsic evaluation by finetuning mBART on the collected data. We find that sentences collected via GIFs are indeed of higher quality.
△ Less
Submitted 12 June, 2021;
originally announced June 2021.
-
On the Sum of Correlated Squared $κ-μ$ Shadowed Random Variables and its Application to Performance Analysis of MRC
Authors:
Manav R. Bhatnagar
Abstract:
In this paper, we study the statistical characterization of the sum of the squared $κ-μ$ shadowed random variables with correlated shadowing components. The probability density function (PDF) of this sum is obtained in the form of a power series. The derived PDF is utilized for obtaining the performance results of the maximal ratio combining (MRC) scheme over correlated $κ-μ$ shadowed fading chann…
▽ More
In this paper, we study the statistical characterization of the sum of the squared $κ-μ$ shadowed random variables with correlated shadowing components. The probability density function (PDF) of this sum is obtained in the form of a power series. The derived PDF is utilized for obtaining the performance results of the maximal ratio combining (MRC) scheme over correlated $κ-μ$ shadowed fading channels. First, we derive the moment generating function (MGF) of the received signal-to-noise ratio of the MRC receiver. By using the derived MGF expression, the analytical diversity order is obtained; it is deduced on the basis of this analysis that the diversity of the MRC receiver over correlated $κ-μ$ shadowed channels depends upon the number of diversity branches and $μ$ parameter. Further, the analytical average bit error rate of the MRC scheme is also derived, which is applicable for $M$-PSK and $M$-QAM constellations. The Shannon capacity of the correlated $κ-μ$ shadowed channels is also derived in the form of the Meijer-G function.
△ Less
Submitted 5 September, 2014;
originally announced September 2014.
-
On the Capacity of CSI Based Transmission Link Selection in Decode-and-Forward Cooperative System
Authors:
Manav R. Bhatnagar
Abstract:
In this paper, we study the problem of best transmission link selection in a decode-and-forward (DF) cooperative system from capacity point of view. The transmission link can be a cooperative (via a relay) or direct link between the source and destination nodes. In a two-hop DF system with multiple relays and a direct link in between the source and destination, the transmission link selection can…
▽ More
In this paper, we study the problem of best transmission link selection in a decode-and-forward (DF) cooperative system from capacity point of view. The transmission link can be a cooperative (via a relay) or direct link between the source and destination nodes. In a two-hop DF system with multiple relays and a direct link in between the source and destination, the transmission link selection can be performed based on full or partial channel state information (CSI) of all links involved in cooperation. We derive analytical ergodic capacity of full and partial CSI based path selection schemes in the DF cooperative system. Further, the full and partial CSI based link selection schemes are compared with help of these expressions.
△ Less
Submitted 4 July, 2014;
originally announced July 2014.
-
On the Capacity of Decode-and-Forward Relaying over Rician Fading Channels
Authors:
Manav R. Bhatnagar
Abstract:
In this letter, we derive the probability density function (PDF) and cumulative distribution function (CDF) of the minimum of two non-central Chi-square random variables with two degrees of freedom in terms of power series. With the help of the derived PDF and CDF, we obtain the exact ergodic capacity of the following adaptive protocols in a decode-and-forward (DF) cooperative system over dissimil…
▽ More
In this letter, we derive the probability density function (PDF) and cumulative distribution function (CDF) of the minimum of two non-central Chi-square random variables with two degrees of freedom in terms of power series. With the help of the derived PDF and CDF, we obtain the exact ergodic capacity of the following adaptive protocols in a decode-and-forward (DF) cooperative system over dissimilar Rician fading channels: (i) constant power with optimal rate adaptation; (ii) optimal simultaneous power and rate adaptation; (iii) channel inversion with fixed rate. By using the analytical expressions of the capacity, it is observed that the optimal power and rate adaptation provides better capacity than the optimal rate adaptation with constant power from low to moderate signal-to-noise ratio values over dissimilar Rician fading channels. Despite low complexity, the channel inversion based adaptive transmission is shown to suffer from significant loss in capacity as compared to the other adaptive transmission based techniques over DF Rician channels.
△ Less
Submitted 3 July, 2014;
originally announced July 2014.
-
Performance Analysis of Two-Way AF MIMO Relaying of OSTBCs with Imperfect Channel Gains
Authors:
Arti M. K.,
Manav R. Bhatnagar
Abstract:
In this paper, we consider the relaying of orthogonal space time block codes (OSTBCs) in a two-way amplify-and-forward (AF) multiple-input multiple-output (MIMO) relay system with estimated channel state information (CSI). A simple four phase protocol is used for training and OSTBC data transmission. Decoding of OSTBC data at a user terminal is performed by replacing the exact CSI by the estimated…
▽ More
In this paper, we consider the relaying of orthogonal space time block codes (OSTBCs) in a two-way amplify-and-forward (AF) multiple-input multiple-output (MIMO) relay system with estimated channel state information (CSI). A simple four phase protocol is used for training and OSTBC data transmission. Decoding of OSTBC data at a user terminal is performed by replacing the exact CSI by the estimated CSI, in a maximum likelihood decoder. Tight approximations for the moment generating function (m.g.f.) of the received signal-to-noise ratio at a user is derived under Rayleigh fading by ignoring the higher order noise terms. Analytical average error performance of the considered cooperative scheme is derived by using the m.g.f. expression. Moreover, the analytical diversity order of the considered scheme is also obtained for certain system configurations. It is shown by simulations and analysis that the channel estimation does not affect the diversity order of the OSTBC based two-way AF MIMO relay system.
△ Less
Submitted 3 July, 2014;
originally announced July 2014.
-
Exploratory Model Building
Authors:
Raj Bhatnagar
Abstract:
Some instances of creative thinking require an agent to build and test hypothetical theories. Such a reasoner needs to explore the space of not only those situations that have occurred in the past, but also those that are rationally conceivable. In this paper we present a formalism for exploring the space of conceivable situation-models for those domains in which the knowledge is primarily proba…
▽ More
Some instances of creative thinking require an agent to build and test hypothetical theories. Such a reasoner needs to explore the space of not only those situations that have occurred in the past, but also those that are rationally conceivable. In this paper we present a formalism for exploring the space of conceivable situation-models for those domains in which the knowledge is primarily probabilistic in nature. The formalism seeks to construct consistent, minimal, and desirable situation-descriptions by selecting suitable domain-attributes and dependency relationships from the available domain knowledge.
△ Less
Submitted 27 February, 2013;
originally announced February 2013.
-
Performance Analysis of Decode-and-Forward Relaying in Gamma-Gamma Fading Channels
Authors:
Manav R. Bhatnagar
Abstract:
Decode-and-forward (DF) cooperative communication based on free space optical (FSO) links is studied in this letter. We analyze performance of the DF protocol in the FSO links following the Gamma-Gamma distribution. The cumulative distribution function (CDF) and probability density function (PDF) of a random variable containing mixture of the Gamma- Gamma and Gaussian random variables is derived.…
▽ More
Decode-and-forward (DF) cooperative communication based on free space optical (FSO) links is studied in this letter. We analyze performance of the DF protocol in the FSO links following the Gamma-Gamma distribution. The cumulative distribution function (CDF) and probability density function (PDF) of a random variable containing mixture of the Gamma- Gamma and Gaussian random variables is derived. By using the derived CDF and PDF, average bit error rate of the DF relaying is obtained.
△ Less
Submitted 2 May, 2012;
originally announced May 2012.
-
Comparing Soft Computing Techniques For Early Stage Software Development Effort Estimations
Authors:
Roheet Bhatnagar,
Mrinal Kanti Ghose
Abstract:
Accurately estimating the software size, cost, effort and schedule is probably the biggest challenge facing software developers today. It has major implications for the management of software development because both the overestimates and underestimates have direct impact for causing damage to software companies. Lot of models have been proposed over the years by various researchers for carrying o…
▽ More
Accurately estimating the software size, cost, effort and schedule is probably the biggest challenge facing software developers today. It has major implications for the management of software development because both the overestimates and underestimates have direct impact for causing damage to software companies. Lot of models have been proposed over the years by various researchers for carrying out effort estimations. Also some of the studies for early stage effort estimations suggest the importance of early estimations. New paradigms offer alternatives to estimate the software development effort, in particular the Computational Intelligence (CI) that exploits mechanisms of interaction between humans and processes domain knowledge with the intention of building intelligent systems (IS). Among IS, Artificial Neural Network and Fuzzy Logic are the two most popular soft computing techniques for software development effort estimation. In this paper neural network models and Mamdani FIS model have been used to predict the early stage effort estimations using the student dataset. It has been found that Mamdani FIS was able to predict the early stage efforts more efficiently in comparison to the neural network models based models.
△ Less
Submitted 28 April, 2012;
originally announced April 2012.
-
Decode-and-Forward Based Differential Modulation for Cooperative Communication System with Unitary and Non-Unitary Constellations
Authors:
Manav R. Bhatnagar
Abstract:
In this paper, we derive a maximum likelihood (ML) decoder of the differential data in a decode-and-forward (DF) based cooperative communication system utilizing uncoded transmissions. This decoder is applicable to complex-valued unitary and non-unitary constellations suitable for differential modulation. The ML decoder helps in improving the diversity of the DF based differential cooperative syst…
▽ More
In this paper, we derive a maximum likelihood (ML) decoder of the differential data in a decode-and-forward (DF) based cooperative communication system utilizing uncoded transmissions. This decoder is applicable to complex-valued unitary and non-unitary constellations suitable for differential modulation. The ML decoder helps in improving the diversity of the DF based differential cooperative system using an erroneous relaying node. We also derive a piecewise linear (PL) decoder of the differential data transmitted in the DF based cooperative system. The proposed PL decoder significantly reduces the decoding complexity as compared to the proposed ML decoder without any significant degradation in the receiver performance. Existing ML and PL decoders of the differentially modulated uncoded data in the DF based cooperative communication system are only applicable to binary modulated signals like binary phase shift keying (BPSK) and binary frequency shift keying (BFSK), whereas, the proposed decoders are applicable to complex-valued unitary and non-unitary constellations suitable for differential modulation under uncoded transmissions. We derive a closedform expression of the uncoded average symbol error rate (SER) of the proposed PL decoder with M-PSK constellation in a cooperative communication system with a single relay and one source-destination pair. An approximate average SER by ignoring higher order noise terms is also derived for this set-up. It is analytically shown on the basis of the derived approximate SER that the proposed PL decoder provides full diversity of second order. In addition, we also derive approximate SER of the differential DF system with multiple relays at asymptotically high signal-to-noise ratio of the source-relay links.
△ Less
Submitted 11 April, 2012;
originally announced April 2012.
