This tutorial targets researchers and practitioners who are interested in AI and ML technologies for structural information extraction (IE) from unstructured textual sources. Particularly, this tutorial will provide audience with a systematic introduction to recent advances of IE, by answering several important research questions. These questions include (i) how to develop an robust IE system from noisy, insufficient training data, while ensuring the reliability of its prediction? (ii) how to foster the generalizability of IE through enhancing the system’s cross-lingual, cross-domain, cross-task and cross-modal transferability? (iii) how to precisely support extracting structural information with extremely fine-grained, diverse and boundless labels? (iv) how to further improve IE by leveraging indirect supervision from other NLP tasks, such as NLI, QA or summarization, and pre-trained language models? (v) how to acquire knowledge to guide the inference of IE systems? We will discuss several lines of frontier research that tackle those challenges, and will conclude the tutorial by outlining directions for further investigation.
more »
« less
This content will become publicly available on January 1, 2024
Indirectly Supervised Natural Language Processing
This tutorial targets researchers and practitioners who are interested in ML technologies for NLP from indirect supervision. In particular, we will present a diverse thread of indirect supervision studies that try to answer the following questions: (i) when and how can we provide supervision for a target task T, if all we have is data that corresponds to a “related” task T′? (ii) humans do not use exhaustive supervision; they rely on occasional feedback, and learn from incidental signals from various sources; how can we effectively incorporate such supervision in machine learning? (iii) how can we leverage multi-modal supervision to help NLP? To the end, we will discuss several lines of research that address those challenges, including (i) indirect supervision from T ′ that handles T with outputs spanning from a moderate size to an open space, (ii) the use of sparsely occurring and incidental signals, such as partial labels, noisy labels, knowledge-based constraints, and cross-domain or cross-task annotations—all having statistical associations with the task, (iii) principled ways to measure and understand why these incidental signals can contribute to our target tasks, and (iv) indirect supervision from vision-language signals. We will conclude the tutorial by outlining directions for further investigation.
more »
« less
- Award ID(s):
- 2105329
- NSF-PAR ID:
- 10440670
- Date Published:
- Journal Name:
- Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 6: Tutorial Abstracts)
- Page Range / eLocation ID:
- 32 to 40
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
We explore the effect of auxiliary labels in improving the classification accuracy of wearable sensor-based human activity recognition (HAR) systems, which are primarily trained with the supervision of the activity labels (e.g. running, walking, jumping). Supplemental meta-data are often available during the data collection process such as body positions of the wearable sensors, subjects' demographic information (e.g. gender, age), and the type of wearable used (e.g. smartphone, smart-watch). This information, while not directly related to the activity classification task, can nonetheless provide auxiliary supervision and has the potential to significantly improve the HAR accuracy by providing extra guidance on how to handle the introduced sample heterogeneity from the change in domains (i.e positions, persons, or sensors), especially in the presence of limited activity labels. However, integrating such meta-data information in the classification pipeline is non-trivial - (i) the complex interaction between the activity and domain label space is hard to capture with a simple multi-task and/or adversarial learning setup, (ii) meta-data and activity labels might not be simultaneously available for all collected samples. To address these issues, we propose a novel framework Conditional Domain Embeddings (CoDEm). From the available unlabeled raw samples and their domain meta-data, we first learn a set of domain embeddings using a contrastive learning methodology to handle inter-domain variability and inter-domain similarity. To classify the activities, CoDEm then learns the label embeddings in a contrastive fashion, conditioned on domain embeddings with a novel attention mechanism, enforcing the model to learn the complex domain-activity relationships. We extensively evaluate CoDEm in three benchmark datasets against a number of multi-task and adversarial learning baselines and achieve state-of-the-art performance in each avenue.more » « less
-
Waleed Khalil (Ed.)The increasing performance demanded by emerging wireless communication standards motivates the development of various techniques devoted to improving the efficiency of power amplifiers (PA) because this is one of the most power-demanding blocks in RF transceivers. Power-amplifier efficiency is proportional to the ratio of the average voltage delivered by the PA to the voltage level of the PA's power supply. Efficiency is affected by the peak-to-average ratio of the transmitted signal. The envelope tracking modulator maximizes this ratio, correlating the PA's power supply with the envelope of its output signal. Efficient modulators must satisfy certain critical conditions: i) it must be very agile to track the amplitude variations of PA's output voltage; ii) it must reduce the timing mismatch between the PA modulator's supply and PA output waveform envelope to optimize power efficiency and avoid PA saturation, and iii) the envelope tracking modulator must be highly power efficient. This paper reviews several relevant envelope tracking techniques. Hybrid modulators consisting of switching regulators and linear amplifiers have become mainstream envelope tracking systems for wideband applications, in which linear amplifiers complement the functionality of highly efficient but narrow bandwidth switching modulators. Replacements for linear amplifiers include a combination of power-efficient ADC and DACs that provide very agile feedback, increasing the system's slew rate, which allows the modulator to track faster envelope signals. Multi-level switching is another relevant approach utilizing multiple switching voltages to reduce current ripples and enable the use of wider bandwidth switching regulators with high power efficiency. The use of multiple inductors is another interesting approach. Multi-phase switching techniques utilize multiple switching stages in a time-interleaved manner to extend the switching modulator's bandwidth. A slow buck converter can be combined with a fast buck converter and optimized for different switching frequencies; this architecture covers the signal envelope's low- and high-frequency components. The approaches mentioned use switching modulators with analog feedback controllers (Pulse-width modulation [PWM] or hysteretic). However, an alternative approach is prediction-based digital feedforward control. This tutorial discusses all of these approaches.more » « less
-
Current studies of bias in NLP rely mainly on identifying (unwanted or negative) bias towards a specific demographic group. While this has led to progress recognizing and mitigating negative bias, and having a clear notion of the targeted group is necessary, it is not always practical. In this work we extrapolate to a broader notion of bias, rooted in social science and psychology literature. We move towards predicting interpersonal group relationship (IGR) - modeling the relationship between the speaker and the target in an utterance - using fine-grained interpersonal emotions as an anchor. We build and release a dataset of English tweets by US Congress members annotated for interpersonal emotion - the first of its kind, and ‘found supervision’ for IGR labels; our analyses show that subtle emotional signals are indicative of different biases. While humans can perform better than chance at identifying IGR given an utterance, we show that neural models perform much better; furthermore, a shared encoding between IGR and interpersonal perceived emotion enabled performance gains in both tasks.more » « less
-
This tutorial will introduce our Accessibility Learning Labs (ALL). The objectives of this collaborative project with The National Technical Institute for the Deaf (NTID) are to both inform participants about foundational topics in accessibility and to demonstrate the importance of creating accessible software. The labs enable easy classroom inclusion by providing instructors all necessary materials including lecture and activity slides and videos. Each lab addresses an accessibility issue and contains: I) Relevant background information on the examined issue II) An example web-based application containing the accessibility problem III) A process to emulate this accessibility problem IV) Details about how to repair the problem from a technical perspective V) Incidents from people who encountered this accessibility issue and how it has impacted their life. The labs may be easily integrated into a wide variety of curriculum at high schools (9-12), and in undergraduate and graduate courses. The labs will be easily adoptable due to their selfcontained nature and their inclusion of all necessary instructional material (e.g., slides, quizzes, etc.). No special software is required to use any portion of the labs since they are web-based and are able to run on any computer with a reasonably recent web browser. There are currently four available labs on the topics of: Colorblindness, Hearing, Blindness and Dexterity. Material is available on our website: http://all.rit.edu This tutorial will provide an overview of the created labs and usage instructions and information for adaptors.more » « less
