Papers by Ernie Chang
Programmable Annotation with Diversed Heuristics and Data Denoising (2022.coling-1)
Copied to clipboard
| Challenge: | Neural natural language generation and understanding models require massive amounts of annotated data to be competitive. |
| Approach: | They propose a data programming framework that can jointly construct labeled data for language generation and understanding tasks by allowing annotators to modify an automatically-inferred alignment rule set between sequence labels and text. |
| Outcome: | The proposed framework generates high-quality data within a 1.48 BLEU and 6.42 slot F1 of 100% human-labeled data with just 100 labeled data samples outperforming benchmark annotation frameworks and other semi-supervised approaches. |
Breaking Down Power Barriers in On-Device Streaming ASR: Insights and Solutions (2025.naacl-industry)
Copied to clipboard
Yang Li, Yuan Shangguan, Yuhao Wang, Liangzhen Lai, Ernie Chang, Changsheng Zhao, Yangyang Shi, Vikas Chandra
| Challenge: | Streaming automatic speech recognition models use high power consumption to improve usability and accuracy. |
| Approach: | They propose to optimize on-device speech recognition models by adjusting component energy sensitivities based on their specific energy sensitities to reduce power consumption. |
| Outcome: | The proposed approach achieves up to 47% lower energy usage while preserving comparable model accuracy and improving real-time performance compared to leading methods. |
On Training Instance Selection for Few-Shot Neural Text Generation (2021.acl-short)
Copied to clipboard
| Challenge: | Pretraining large neural networks with a language modeling objective has led to dramatic improvements in text generation. |
| Approach: | They propose a selection strategy to select few-shot training instances based on unlabeled data to identify the most worthwhile data points that should be annotated under some budget of labeling cost. |
| Outcome: | The proposed strategy outperforms random sampling on three text generation tasks. |
Neural Data-to-Text Generation via Jointly Learning the Segmentation and Correspondence (2020.acl-main)
Copied to clipboard
| Challenge: | Recent neural attention models conflate all steps into a single end-to-end system and simplify training process. |
| Approach: | They propose to explicitly segment target text into fragment units and align them with their data correspondences. |
| Outcome: | The proposed model outperforms neural attention models on E2E and WebNLG benchmarks. |
DART: A Lightweight Quality-Suggestive Data-to-Text Annotation Tool (2020.coling-demos)
Copied to clipboard
| Challenge: | Neural data-to-text generation systems require large-scale labeled data to generate sentences. |
| Approach: | They propose to create an interactive annotation tool that iteratively analyzes annotated structured data to better sample unlabeled data. |
| Outcome: | The proposed tool reduces the number of annotations needed with active learning and automatically suggests relevant labels. |
Revisiting Sample Size Determination in Natural Language Understanding (2023.findings-acl)
Copied to clipboard
Ernie Chang, Muhammad Hassan Rashid, Pin-Jie Lin, Changsheng Zhao, Vera Demberg, Yangyang Shi, Vikas Chandra
| Challenge: | Recent work has sought to reduce the annotation costs through the use of active learning and data sampling. |
| Approach: | They propose to estimate the training sample size needed to achieve a targeted model performance based on small amount of training samples. |
| Outcome: | The proposed approach predicts model performance within a small margin of mean absolute error (0.9%) with only 10% data. |
Towards Zero-Shot Multilingual Transfer for Code-Switched Responses (2023.acl-long)
Copied to clipboard
| Challenge: | Recent task-oriented dialog systems have had great success building English-based personal assistants, but extending these systems to a global audience may take tremendous efforts. |
| Approach: | They propose a framework that allows for efficient transfer by learning task-specific representations and encapsulating source and target language representations. |
| Outcome: | The proposed framework is able to successfully transfer language knowledge even when the target language corpus is limited. |
Scaling Parameter-Constrained Language Models with Quality Data (2024.emnlp-industry)
Copied to clipboard
Ernie Chang, Matteo Paltenghi, Yang Li, Pin-Jie Lin, Changsheng Zhao, Patrick Huber, Zechun Liu, Rastislav Rabatin, Yangyang Shi, Vikas Chandra
| Challenge: | Scaling laws in language modeling quantify training loss as a function of dataset size and model parameters, but neglect the critical role of data quality in model generalization. |
| Approach: | They propose to use effective training tokens as a combination of text diversity and syntheticity as measured by a teacher model to calculate scaling laws. |
| Outcome: | The proposed term effective training tokens is a combination of two readily-computed indicators of text diversity and syntheticity as measured by a teacher model. |
Does the Order of Training Samples Matter? Improving Neural Data-to-Text Generation with Curriculum Learning (2021.eacl-main)
Copied to clipboard
| Challenge: | Recent advances in data-to-text generation have been focused on curriculum learning, which is a process of presenting training data in a specific order, starting from easy examples and moving on to more difficult ones, as the learner becomes more competent. |
| Approach: | They propose to use a curriculum learning process to change the order of training samples in a model based on the model's competence to improve model performance and convergence speed. |
| Outcome: | The proposed model shows faster convergence speed and reduced training time by 38.7% and performance by 4.84 BLEU. |
MovieChats: Chat like Humans in a Closed Domain (2020.emnlp-main)
Copied to clipboard
| Challenge: | Currently, open-domain chatbots are far from satisfactory. |
| Approach: | They propose a unified, readily scalable neural approach which reconciles all subtasks like intent prediction and knowledge retrieval. |
| Outcome: | The proposed approach outperforms commercial systems replying on complex rules on static and interactive tests and shows that the results are remarkably good. |
Neural Data-to-Text Generation with LM-based Text Augmentation (2021.eacl-main)
Copied to clipboard
| Challenge: | Neural data-to-text generation is a difficult task for many new applications because of a lack of training data. |
| Approach: | They propose a few-shot approach that augments the data available for training by generating new text samples based on replacing specific values by alternative ones from the same category and pairing the new text with data samples. |
| Outcome: | The proposed approach outperforms fully supervised sequence-to-sequence models with less than 10% of the training set on both datasets. |
Few-Shot Pidgin Text Adaptation via Contrastive Fine-Tuning (2022.coling-1)
Copied to clipboard
| Challenge: | Currently, low resource languages are not supported by proper translation systems or parallel corpus. |
| Approach: | They propose to fine-tune the pretrained language models to generate utterances in English-to-Pidgin by leveraging the proximity of the source and target languages and using positive and negative examples in constrastive training objectives. |
| Outcome: | The proposed method is sufficient to generate utterances in English-to-Pidgin, which are two closely-related languages. |
Jointly Improving Language Understanding and Generation with Quality-Weighted Weak Supervision of Automatic Labeling (2021.eacl-main)
Copied to clipboard
| Challenge: | Neural natural language generation and understanding models are data-hungry and require massive amounts of annotated data to be competitive. |
| Approach: | They propose a framework that automatically synthesizes weak labels from large-scale weakly-labeled data with a fine-tuned GPT-2 and adapts parameter updates to the models according to the estimated label-quality. |
| Outcome: | The proposed framework outperforms benchmark systems on the E2E and Weather datasets when 100% of the training data is used. |
Improving Language Generation from Feature-Rich Tree-Structured Data with Relational Graph Convolutional Encoders (D19-63)
Copied to clipboard
| Challenge: | The goal of the multilingual surface realization shared task is to generate fluent text from UD structures. |
| Approach: | They propose to use a graph convolutional network to encode the dependency trees given as input. |
| Outcome: | The proposed system achieves the third rank without data augmentation techniques or additional components. |
Self-Vocabularizing Training for Neural Machine Translation (2025.naacl-srw)
Copied to clipboard
| Challenge: | Past vocabulary learning techniques identify relevant vocabulary before training, relying on corpus statistics or frequency counts without considering contextual information or the model's ability to represent it. |
| Approach: | They propose a method that self-vocabularizes a smaller, more optimal vocabulary by pairing source sentences with the model's predictions to define a new vocabulary. |
| Outcome: | The proposed method produces a 1.49 BLEU improvement in the simulated model and an increase in unique token usage and a 6–8% reduction in vocabulary size. |
Target-Aware Language Modeling via Granular Data Sampling (2024.emnlp-main)
Copied to clipboard
Ernie Chang, Pin-Jie Lin, Yang Li, Changsheng Zhao, Daeil Kim, Rastislav Rabatin, Zechun Liu, Yangyang Shi, Vikas Chandra
| Challenge: | Language model pretraining is the cornerstone of universal language models (LMs), creating generalpurpose representations to excel across a variety of downstream tasks. |
| Approach: | They propose to use multi-granular tokens to sample large-scale language models for domain-specific use cases. |
| Outcome: | The proposed model outperforms random sampled samples on eight benchmarks with 1% of the data and performs on par with the full RefinedWeb data. |
LLM-QAT: Data-Free Quantization Aware Training for Large Language Models (2024.findings-acl)
Copied to clipboard
Zechun Liu, Barlas Oguz, Changsheng Zhao, Ernie Chang, Pierre Stock, Yashar Mehdad, Yangyang Shi, Raghuraman Krishnamoorthi, Vikas Chandra
| Challenge: | Several post-training quantization methods have been shown to perform well down to 8-bits. |
| Approach: | They propose a data-free distillation method that leverages generations produced by the pre-trained model to quantize any generative model independent of its training data. |
| Outcome: | The proposed method outperforms SoTA PTQ and LLaMA models at low bit precision. |
Improving Zero-Shot Multilingual Text Generation via Iterative Distillation (2022.coling-1)
Copied to clipboard
| Challenge: | Existing approaches to generalize multilingual dialogue systems to multilingual settings often make assumptions about data availability. |
| Approach: | They propose to transfer inductive biases for target languages learned by pretrained teacher models to student models via sequence-level knowledge distillation. |
| Outcome: | The proposed method performs well on the multiATIS++ benchmark, and is comparable to human annotations in both slot F1 and intent accuracy. |
AutoMixer: Checkpoint Artifacts as Automatic Data Mixers (2025.acl-long)
Copied to clipboard
| Challenge: | In language model training, it is difficult to obtain the right data mixtures for various tasks as the relationship between data and tasks is difficult. |
| Approach: | They propose to identify checkpoint models based on their respective capabilities and leverage them as data mixers by using their aggregated first-order influence approximation over source data. |
| Outcome: | The proposed framework shows significant improvements on eight reasoning benchmarks, with accuracy increases of up to 1.93%. |
Logic-Guided Message Generation from Raw Real-Time Sensor Data (2022.lrec-1)
Copied to clipboard
| Challenge: | Developing a natural language generation model to enable human pilots to communicate with drones is challenging because of its redundant nature and diversity. |
| Approach: | They propose a corpus for a specific domain that instantiates these properties by combining sensor data with text. |
| Outcome: | The proposed model can alert the human pilot of the system state and environment in preparation of handover of control. |
IMPACT: Importance-Aware Activation Space Reconstruction (2026.acl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) achieve strong performance across domains but remain difficult to deploy in resource-constrained environments due to their massive size. |
| Approach: | They propose an importance-aware activation reconstruction framework that links compression to its effect on model performance. |
| Outcome: | Experiments show that IMPACT reduces model size by 55.4% while maintaining accuracy comparable to or better than state-of-the-art models. |
A Few Thousand Translations Go a Long Way! Leveraging Pre-trained Models for African News Translation (2022.naacl-main)
Copied to clipboard
David Adelani, Jesujoba Alabi, Angela Fan, Julia Kreutzer, Xiaoyu Shen, Machel Reid, Dana Ruiter, Dietrich Klakow, Peter Nabende, Ernie Chang, Tajuddeen Gwadabe, Freshia Sackey, Bonaventure F. P. Dossou, Chris Emezue, Colin Leong, Michael Beukman, Shamsuddeen Muhammad, Guyo Jarso, Oreen Yousuf, Andre Niyongabo Rubungo, Gilles Hacheme, Eric Peter Wairagala, Muhammad Umair Nasir, Benjamin Ajibade, Tunde Ajayi, Yvonne Gitau, Jade Abbott, Mohamed Ahmed, Millicent Ochieng, Anuoluwapo Aremu, Perez Ogayo, Jonathan Mukiibi, Fatoumata Ouoba Kabore, Godson Kalipe, Derguene Mbaye, Allahsera Auguste Tapo, Victoire Memdjokam Koagne, Edwin Munkoh-Buabeng, Valencia Wagner, Idris Abdulmumin, Ayodele Awokoya, Happy Buzaaba, Blessing Sibanda, Andiswa Bukula, Sam Manthalu
| Challenge: | Low-resource languages are left out of large-scale pretraining datasets . authors explore how to leverage existing pre-trained models to create low-resourced translation systems for 16 African languages. |
| Approach: | They investigate how large-scale pre-trained models can be used to create low-resource translation systems for 16 African languages. |
| Outcome: | The proposed models can translate between hundreds of languages even though there is little parallel data available for training. |