본문 바로가기
KERD
논문KCI 등재

연속학습 기반의 대화형 AI agent

A Study on Conversational AI Agent based on Continual Learning

정옥란|박채림|유소엽
한국컴퓨터정보학회 (The Korean Society Of Computer And Information)|2023.01|한국컴퓨터정보학회논문지 (Journal of The Korea Society of Computer and Information)|제28권 제1호|pp. 27-38|DOI https://doi.org/10.9708/jksci.2023.28.01.027

0

조회

0

다운로드

4

인용

0

참고문헌

키워드

챗봇대화 시스템ChatbotsKnowledge graphContinual learning지식 그래프대화형 AI 에이전트연속학습Dialogue systemConversational AI agent
주제분류컴퓨터학

초록

본 논문에서는 시간의 흐름에 따라 새로운 데이터를 지속적으로 학습하고 성장할 수 있는 연속학습 기반 대화형 AI 에이전트를 제안한다. 연속학습 기반 대화형 AI 에이전트는 태스크 관리자(Task Manager), 사용자 속성 추출(User Attribute Extraction), 자동 확장 지식 그래프(Auto-growing Knowledge Graph), 크게 3가지 요소로 구성된다. 태스크 관리자는 사용자와의 대화에서 새로운 데이터를 발견하면 이전에 학습한 지식을 통해 새로운 태스크를 생성한다. 사용자 특성 추출 모델은 새로운 태스크에서 사용자의 특성을 추출하고, 자동 확장 지식 그래프는 새로운 외부 지식을지속적으로 학습할 수 있도록 한다. 한정된 데이터셋을 기반으로 학습된 기존 대화형 AI 에이전트와 달리, 본 논문에서 제안하는 방법은 지속적인 사용자의 특성과 지식 학습을 기반으로 대화를 가능하게 한다. 연속학습 기술이 적용된 대화형 AI 에이전트는 사용자와의 대화가 축적될수록개인 맞춤형 대응이 가능하며, 새로운 지식에도 대응이 가능하다. 본 논문에서는 시간에 따른 대화 생성 모델의 성능 변화 실험을 통해 제안하는 방법의 가능성을 검증한다.

Abstract

In this paper, we propose a conversational AI agent based on continual learning that can continuously learn and grow with new data over time. A continual learning-based conversational AI agent consists of three main components: Task manager, User attribute extraction, and Auto-growing knowledge graph. When a task manager finds new data during a conversation with a user, it creates a new task with previously learned knowledge. The user attribute extraction model extracts the user’s characteristics from the new task, and the auto-growing knowledge graph continuously learns the new external knowledge. Unlike the existing conversational AI agents that learned based on a limited dataset, our proposed method enables conversations based on continuous user attribute learning and knowledge learning. A conversational AI agent with continual learning technology can respond personally as conversations with users accumulate. And it can respond to new knowledge continuously. This paper validate the possibility of our proposed method through experiments on performance changes in dialogue generation models over time.

참고문헌 (52)

  1. ROBERT DALE, "The return of the chatbots", Natural Language Engineering, 22(5), pp.811~817 (2016), Cambridge University Press (CUP)[DOI]
  2. S. Diederich, "On the Design of and Interaction with Conversational Agents: An Organizing and Assessing Review of Human-Computer Interaction Research", Journal of the Association for Information Systems, 23(1), pp.96~138 (2022), Association for Information Systems[DOI]
  3. S. Mazumder, "Lifelong and Interactive Learning of Factual Knowledge in Dialogues", 20th Annual Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL), pp.21~31 (2019), TH Royal Inst Technol
  4. A. Chaudhry, "Using Hindsight to Anchor Past Knowledge in Continual Learning", 35th AAAI Conference on Artificial Intelligence, 11th Symposium on Educational Advances in Artificial Intelligence, pp.2021[DOI]
  5. Timothée Lesort, "Continual learning for robotics: Definition, framework, learning strategies, opportunities and challenges", Information Fusion, 58, pp.52~68 (2020), Elsevier BV[DOI]
  6. Matthias Delange, "A continual learning survey: Defying forgetting in classification tasks", IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(7), pp.3366~3385 (2021), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
  9. H. Shin, "Continual Learning with Deep Generative Replay", 31st Annual Conference on Neural Information Processing Systems (NIPS) (2017)
  10. C. S. Wu, "Memory Replay GANs: learning to generate images from new categories without forgetting", 32nd Conference on Neural Information Processing Systems (NIPS) (2018)
  12. S. A. Rebuffi, "iCaRL: Incremental Classifier and Representation Learning", 30th IEEE, 2017, 5533~5542
  13. A. Rios, "Closed-Loop Memory GAN for Continual Learning", 28th International Joint Conference on Artificial Intelligence, pp.3332~3338 (2019)
  14. C. V. Nguyen, "Variational continual learning", ICLR 2018 (2017)
  15. J. Kirkpatricka, "Overcoming catastrophic forgetting in neural networks", Proceedings of the National Academy of Sciences of the United States of America, 114(13), pp.3521~3526 (2017)
  16. F. Zenke, "Continual Learning Through Synaptic Intelligence", 34th International Conference on Machine Learning (2017)
  17. Davide Maltoni, "Continuous learning in single-incremental-task scenarios", Neural Networks, 116, pp.56~73 (2019), Elsevier BV[DOI]
  18. Z. Z. Li, "Learning Without Forgetting", 14th European Conference on Computer Vision(ECCV), pp.614~629 (2016)
  20. R. Kemker, "Fearnet: Brain-inspired model for incremental learning", ICLR 2018 (2017)
  21. V. Lomonaco, "CORe50: a New Dataset and Benchmark for Continuous Object Recognition", Journal, 78, pp.17~26 (2017)
  22. A. Chaudhry, "Continual learning with tiny episodic memories", Journal (2019)
  23. D. Lopez-Paz, "Gradient Episodic Memory for Continual Learning", 31st Annual Conference on Neural Information Processing Systems (NIPS) (2017)
  24. A. Chaudhry, "Efficient lifelong learning with a-gem", ICLR 2019 (2018)
  25. M. Mancini, "Adding New Tasks to a Single Network with Weight Transformations Using Binary Masks", 15th European Conference on Computer Vision (ECCV), pp.80~189 (2019)
  26. J. Xu, "Reinforced Continual Learning", 2nd Conference on Neural Information Processing Systems (NIPS) (2018)
  27. Amir Rosenfeld, "Incremental Learning Through Deep Adaptation", IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(3), pp.651~663 (2020), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
  28. A. Mallya, "PackNet: Adding Multiple Tasks to a Single Network by Iterative Pruning", 31st IEEE, 2018, 7765~7773
  29. J. Serra, "Overcoming Catastrophic Forgetting with Hard Attention to the Task", 35th International Conference on Machine Learning (ICML) (2018)
  30. E. Adamopoulou, "An overview of chatbot technology", Journal, pp.373~383 (2020)
  31. R. Takanobu, "Is your goal-oriented dialog model performing really well? empirical analysis of system-wise evaluation", Journal (2020)[DOI]
  33. Joseph Weizenbaum, "ELIZA — a computer program for the study of natural language communication between man and machine", Communications of the ACM, 9(1), pp.36~45 (1983), Association for Computing Machinery (ACM)[DOI]
  34. Kenneth Mark Colby, "Turing-like indistinguishability tests for the validation of a computer simulation of paranoid processes", Artificial Intelligence, 3, pp.199~221 (1972), Elsevier BV[DOI]
  35. B. Liu, "Lifelong and Continual Learning Dialogue Systems: Learning during Conversation", 35th AAAI Conference on Artificial Intelligence, 11th Symposium on Educational Advances in Artificial Intelligence, pp.2021[DOI]
  37. F. M. Suchanek, "Yago: a core of semantic knowledge", Journal, pp.697~706 (2007)[DOI]
  38. S. Auer, "DBpedia: A nucleus for a web of open data", 6th International Semantic Web Conference, ASWC 2007), pp.2007[DOI]
  39. George A. Miller, "WordNet: a lexical database for English", Communications of the ACM, 38(11), pp.39~41 (1995), Association for Computing Machinery (ACM)[DOI]
  40. K. Bollacker, "Freebase: a collaboratively created graph database for structuring human knowledge", Proceedings of the 2008 ACM SIGMOD international conference on Management of data, pp.1247~1250 (1995)
  41. Xiaojun Chen, "A review: Knowledge reasoning over knowledge graph", Expert Systems with Applications, 141, pp.112948~ (2020), Elsevier BV[DOI]
  42. SoYeop Yoo, "Automating the expansion of a knowledge graph", Expert Systems with Applications, 141, pp.112965~ (2020), Elsevier BV[DOI]
  43. Yeon-Sun Ahn, "Time-Aware PolarisX: Auto-Growing Knowledge Graph", Cmc-Computers Materials & Continua, 67(3), pp.2695~2708 (2021), Computers, Materials and Continua (Tech Science Press)[DOI]
  45. D. M. Blei, "Latent dirichlet allocation", Journal of machine Learning research, 3, pp.993~1022 (2003)
  47. S. Kim, "Beyond Domain APIs:Task-oriented Conversational Modeling with Unstructured Knowledge Access", 21st Annual Meeting of the Special-Interest-Group-on-Discourse-and-Dialogue (SIGDIAL), pp.278~289 (2020)
  48. M. Eric, "MultiWOZ 2.1: A Consolidated Multi-Domain Dialogue Dataset with State Corrections and State Tracking Baselines", 12th International Conference on Language Resources and Evaluation (LREC), pp.422~428 (2020)
  50. Y. Z. Zhang, "DIALOGPT : Large-Scale Generative Pre-training for Conversational Response Generation", 58th Annual Meeting of the Association-for-Computational-Linguistics (ACL), pp.270~278 (2020), Electr Network
  51. C.-Y. Lin, "Rouge: A package for automatic evaluation of summaries", Text summarization branches out, 74~81
  52. K. Papineni, "BLEU:a method for automatic evaluation of machine translation", 40th Annual Meeting of the Association-for-Computational-Linguist ics, pp.311~318 (2002), Univ Penn

관련 논문AI추천

화행별 템플릿 기반의 지식획득 기법과유전자 프로그래밍을 이용한 문장 생성 기법을 통한대화형 에이전트의 대화 학습임성수, 조성배, 홍진혁한국인지과학회 · 인지과학2005 · 제16권 제4호페르소나 기반 One-shot MBTI 프롬프트 엔지니어링성연식, 이용재, 지영민한국정보처리학회 · 정보처리학회 논문지2025 · 제14권 제8호심층학습을 이용한 대화시스템의 연구방법 분석김희동, 지인영국제언어인문학회 · 인문언어2018 · 제20권 제2호사용자와 실시간으로 감성적 소통이 가능한 한국어 챗봇 시스템 개발백성대, 이민호한국센서학회 · 센서학회지2021 · 제30권 제6호목적지향 대화 시스템을 위한 챗봇 연구권오욱, 김영길, 이경순, 황금하한국정보처리학회 · 정보처리학회 논문지2017 · 제6권 제11호어댑티브 어시스턴트 기술을 활용한 챗봇 시스템 구축 방법 연구신승근한국IT정책경영학회 · 한국IT정책경영학회 논문지2019 · 제11권 제6호

이 간행물의 다른 논문 (23건)

샴 네트워크 기반 전장상황 분석을 위한 추천 학습모델서창호, 이우신, 신유경, 서기원, 진소연, 안종철pp. 1-8선호 자원 필터링을 위한 빅파이브 개인 유형을 사용한 추론 시스템박종현pp. 9-16LSTM 학습모델의 성능 향상을 위한 디지털 신호 필터의 비교연구 (학습 모델 예측 성능 향상에 영향을 미치는 비트코인 시계열 데이터 전처리 및 스무딩 기법 적용)심갑식, 박태진pp. 17-26딥러닝 기반 2차원 자세 추정을 활용한 골프 스윙 유사도 측정 방법론박종혁pp. 39-47가속도-자이로 센서 퓨전을 이용한 위치 추적 시스템 설계 및 구현에 관한 연구강진구pp. 49-54
원문 보기