{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/a-tree-search-algorithm-for-sequence-labeling","title":"A Tree Search Algorithm for Sequence Labeling","arxiv_id":"1804.10911","date":"2018-04-29","proceeding":null,"authors":["Yadi Lao","Jun Xu","Yanyan Lan","Jiafeng Guo","Sheng Gao","Xue-Qi Cheng"],"abstract":"In this paper we propose a novel reinforcement learning based model for\nsequence tagging, referred to as MM-Tag. Inspired by the success and\nmethodology of the AlphaGo Zero, MM-Tag formalizes the problem of sequence\ntagging with a Monte Carlo tree search (MCTS) enhanced Markov decision process\n(MDP) model, in which the time steps correspond to the positions of words in a\nsentence from left to right, and each action corresponds to assign a tag to a\nword. Two long short-term memory networks (LSTM) are used to summarize the past\ntag assignments and words in the sentence. Based on the outputs of LSTMs, the\npolicy for guiding the tag assignment and the value for predicting the whole\ntagging accuracy of the whole sentence are produced. The policy and value are\nthen strengthened with MCTS, which takes the produced raw policy and value as\ninputs, simulates and evaluates the possible tag assignments at the subsequent\npositions, and outputs a better search policy for assigning tags. A\nreinforcement learning algorithm is proposed to train the model parameters. Our\nwork is the first to apply the MCTS enhanced MDP model to the sequence tagging\ntask. We show that MM-Tag can accurately predict the tags thanks to the\nexploratory decision making mechanism introduced by MCTS. Experimental results\nshow based on a chunking benchmark showed that MM-Tag outperformed the\nstate-of-the-art sequence tagging baselines including CRF and CRF with LSTM.","url_abs":"http://arxiv.org/abs/1804.10911v2","url_pdf":"http://arxiv.org/pdf/1804.10911v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"a-tree-search-algorithm-for-sequence-labeling","repo_url":"https://github.com/YadiLao/MM-Tag","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"chunking","task_name":"Chunking"},{"task_slug":"decision-making","task_name":"Decision Making"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"sentence","task_name":"Sentence"},{"task_slug":"tag","task_name":"TAG"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[{"method_slug":"crf","method_name":"CRF"},{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}