{"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/chemception-a-deep-neural-network-with","title":"Chemception: A Deep Neural Network with Minimal Chemistry Knowledge Matches the Performance of Expert-developed QSAR/QSPR Models","arxiv_id":"1706.06689","date":"2017-06-20","proceeding":null,"authors":["Garrett B. Goh","Charles Siegel","Abhinav Vishnu","Nathan O. Hodas","Nathan Baker"],"abstract":"In the last few years, we have seen the transformative impact of deep\nlearning in many applications, particularly in speech recognition and computer\nvision. Inspired by Google's Inception-ResNet deep convolutional neural network\n(CNN) for image classification, we have developed \"Chemception\", a deep CNN for\nthe prediction of chemical properties, using just the images of 2D drawings of\nmolecules. We develop Chemception without providing any additional explicit\nchemistry knowledge, such as basic concepts like periodicity, or advanced\nfeatures like molecular descriptors and fingerprints. We then show how\nChemception can serve as a general-purpose neural network architecture for\npredicting toxicity, activity, and solvation properties when trained on a\nmodest database of 600 to 40,000 compounds. When compared to multi-layer\nperceptron (MLP) deep neural networks trained with ECFP fingerprints,\nChemception slightly outperforms in activity and solvation prediction and\nslightly underperforms in toxicity prediction. Having matched the performance\nof expert-developed QSAR/QSPR deep learning models, our work demonstrates the\nplausibility of using deep neural networks to assist in computational chemistry\nresearch, where the feature engineering process is performed primarily by a\ndeep learning algorithm.","url_abs":"http://arxiv.org/abs/1706.06689v1","url_pdf":"http://arxiv.org/pdf/1706.06689v1.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":"chemception-a-deep-neural-network-with","repo_url":"https://github.com/Abdulk084/Chemception","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"chemception-a-deep-neural-network-with","repo_url":"https://github.com/Bunseki2/DeepL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"computational-chemistry","task_name":"Computational chemistry"},{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"feature-engineering","task_name":"Feature Engineering"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"prediction","task_name":"Prediction"},{"task_slug":"speech-recognition","task_name":"Speech Recognition"},{"task_slug":"image-classification","task_name":"image-classification"},{"task_slug":"speech-recognition-1","task_name":"speech-recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1706.06689","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}