{"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/molding-cnns-for-text-non-linear-non","title":"Molding CNNs for text: non-linear, non-consecutive convolutions","arxiv_id":"1508.04112","date":"2015-08-17","proceeding":"EMNLP 2015 9","authors":["Tao Lei","Regina Barzilay","Tommi Jaakkola"],"abstract":"The success of deep learning often derives from well-chosen operational\nbuilding blocks. In this work, we revise the temporal convolution operation in\nCNNs to better adapt it to text processing. Instead of concatenating word\nrepresentations, we appeal to tensor algebra and use low-rank n-gram tensors to\ndirectly exploit interactions between words already at the convolution stage.\nMoreover, we extend the n-gram convolution to non-consecutive words to\nrecognize patterns with intervening words. Through a combination of low-rank\ntensors, and pattern weighting, we can efficiently evaluate the resulting\nconvolution operation via dynamic programming. We test the resulting\narchitecture on standard sentiment classification and news categorization\ntasks. Our model achieves state-of-the-art performance both in terms of\naccuracy and training speed. For instance, we obtain 51.2% accuracy on the\nfine-grained sentiment classification task.","url_abs":"http://arxiv.org/abs/1508.04112v2","url_pdf":"http://arxiv.org/pdf/1508.04112v2.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":"molding-cnns-for-text-non-linear-non","repo_url":"https://github.com/taolei87/text_convnet","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"molding-cnns-for-text-non-linear-non","repo_url":"https://github.com/taolei87/rcnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"sentiment-analysis","task_name":"Sentiment Analysis"},{"task_slug":"sentiment-classification","task_name":"Sentiment Classification"},{"task_slug":"tensor-algebra","task_name":"tensor algebra"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1508.04112","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}