Chenlei Guo
Chenlei Guo has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Sixing Lu
Amazon Alexa AI, Seattle, WA, USA
2 shared publications - Benjamin Yao
Amazon.com, Seattle, WA, USA
1 shared publication Chengyuan Ma
Amazon Alexa AI,Seattle,WA,USA
1 shared publication- Chengyuan Ma
Amazon Alexa AI, Seattle, WA, USA
1 shared publication - Clint Solomon Mathialagan
Oracle America Inc
1 shared publication - Gustavo Aguilar
Amazon Alexa AI, Seattle, WA, USA
1 shared publication
Latest Publications
- 5 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Computational Intelligence Magazine
- Vol. 19 (2) , 34-48
A Self-Learning Framework for Large-Scale Conversational AI Systems
In the last decade, conversational artificial intelligence (AI) systems have been widely employed to address people’s real-life needs across various different environments and settings. At the same time, users’ expectations of these systems have been on the rise as they expect more contextual and personalized interactions with continuous learning systems, akin to their expectation in human-human interactions. Modular systems constructed as pipelines of machine learning models and trained through supervised learning paradigms often struggle to improve user experience due to the less-than-ideal, slow accuracy improvements they undergo, and the privacy concerns associated with manual annotation. Inspired by how humans learn from their experiences and interactions, this article proposes a comprehensive self-learning framework designed to tackle these challenges for large-scale conversational AI systems, fostering continuous automated learning. The proposed self-learning framework comprises three elements: feedback collection, feedback interpretation, and learning mechanisms. Without the need for annotators in the loop, a self-learning conversational AI system autonomously uses a feedback interpreter to subscribe to, interpret, and utilize user feedback to adapt its behaviors through various learning mechanisms. First, the elements of the self-learning framework are described and then applied to Alexa, a large-scale conversational AI system. Subsequently, this article presents its effectiveness in reducing user-perceived defects. Finally, it explores the implications of self-learning for general AI systems and suggests future directions.
- 5 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Computational Intelligence Magazine
- Vol. 19 (2) , 34-48
A Self-Learning Framework for Large-Scale Conversational AI Systems
In the last decade, conversational artificial intelligence (AI) systems have been widely employed to address people’s real-life needs across various different environments and settings. At the same time, users’ expectations of these systems have been on the rise as they expect more contextual and personalized interactions with continuous learning systems, akin to their expectation in human-human interactions. Modular systems constructed as pipelines of machine learning models and trained through supervised learning paradigms often struggle to improve user experience due to the less-than-ideal, slow accuracy improvements they undergo, and the privacy concerns associated with manual annotation. Inspired by how humans learn from their experiences and interactions, this article proposes a comprehensive self-learning framework designed to tackle these challenges for large-scale conversational AI systems, fostering continuous automated learning. The proposed self-learning framework comprises three elements: feedback collection, feedback interpretation, and learning mechanisms. Without the need for annotators in the loop, a self-learning conversational AI system autonomously uses a feedback interpreter to subscribe to, interpret, and utilize user feedback to adapt its behaviors through various learning mechanisms. First, the elements of the self-learning framework are described and then applied to Alexa, a large-scale conversational AI system. Subsequently, this article presents its effectiveness in reducing user-perceived defects. Finally, it explores the implications of self-learning for general AI systems and suggests future directions.
- Open Access
- 18 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 2256-2260
PersonalTM: Transformer Memory for Personalized Retrieval
The Transformer Memory as a Differentiable Search Index (DSI) has been proposed as a new information retrieval paradigm, which aims to address the limitations of dual-encoder retrieval framework based on the similarity score. The DSI framework outperforms strong baselines by directly generating relevant document identifiers from queries without relying on an explicit index. The memorization power of DSI framework makes it suitable for personalized retrieval tasks. Therefore, we propose a Personal Transformer Memory (PersonalTM) architecture for personalized text retrieval. PersonalTM incorporates user-specific profiles and contextual user click behaviors, and introduces hierarchical loss in the decoding process to align with the hierarchical assignment of document identifier. Additionally, PersonalTM also employs an adapter architecture to improve the scalability for index updates and reduce computation costs, compared to the vanilla DSI. Experiments show that PersonalTM outperforms the DSI baseline, BM25, fine-tuned dual-encoder, and other personalized models in terms of precision at top 1st and 10th positions and Mean Reciprocal Rank (MRR). Specifically, PersonalTM improves [email protected] by 58%, 49%, and 12% compared to BM25, Dual-encoder, and DSI, respectively.
- 18 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 2256-2260
Open AccessPersonalTM: Transformer Memory for Personalized Retrieval
The Transformer Memory as a Differentiable Search Index (DSI) has been proposed as a new information retrieval paradigm, which aims to address the limitations of dual-encoder retrieval framework based on the similarity score. The DSI framework outperforms strong baselines by directly generating relevant document identifiers from queries without relying on an explicit index. The memorization power of DSI framework makes it suitable for personalized retrieval tasks. Therefore, we propose a Personal Transformer Memory (PersonalTM) architecture for personalized text retrieval. PersonalTM incorporates user-specific profiles and contextual user click behaviors, and introduces hierarchical loss in the decoding process to align with the hierarchical assignment of document identifier. Additionally, PersonalTM also employs an adapter architecture to improve the scalability for index updates and reduce computation costs, compared to the vanilla DSI. Experiments show that PersonalTM outperforms the DSI baseline, BM25, fine-tuned dual-encoder, and other personalized models in terms of precision at top 1st and 10th positions and Mean Reciprocal Rank (MRR). Specifically, PersonalTM improves [email protected] by 58%, 49%, and 12% compared to BM25, Dual-encoder, and DSI, respectively.
- Open Access
- 4 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Clicker: Attention-Based Cross-Lingual Commonsense Knowledge Transfer
Recent advances in cross-lingual commonsense reasoning (CSR) are facilitated by the development of multilingual pre-trained models (mPTMs). While mPTMs show the potential to encode commonsense knowledge for different languages, transferring commonsense knowledge learned in large-scale English corpus to other languages is challenging. To address this problem, we propose the attention-based Cross-LIngual Commonsense Knowledge transfER (CLICKER) framework, which minimizes the performance gaps between English and non-English languages in commonsense question-answering tasks. CLICKER effectively improves commonsense reasoning for non-English languages by differentiating non-commonsense knowledge from commonsense knowledge. Experimental results on public benchmarks demonstrate that CLICKER achieves remarkable improvements in the cross-lingual CSR task for languages other than English.
- 4 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Open AccessClicker: Attention-Based Cross-Lingual Commonsense Knowledge Transfer
Recent advances in cross-lingual commonsense reasoning (CSR) are facilitated by the development of multilingual pre-trained models (mPTMs). While mPTMs show the potential to encode commonsense knowledge for different languages, transferring commonsense knowledge learned in large-scale English corpus to other languages is challenging. To address this problem, we propose the attention-based Cross-LIngual Commonsense Knowledge transfER (CLICKER) framework, which minimizes the performance gaps between English and non-English languages in commonsense question-answering tasks. CLICKER effectively improves commonsense reasoning for non-English languages by differentiating non-commonsense knowledge from commonsense knowledge. Experimental results on public benchmarks demonstrate that CLICKER achieves remarkable improvements in the cross-lingual CSR task for languages other than English.