{"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/adaptive-bidirectional-backpropagation","title":"Adaptive Bidirectional Backpropagation: Towards Biologically Plausible Error Signal Transmission in Neural Networks","arxiv_id":"1702.07097","date":"2017-02-23","proceeding":null,"authors":["Hongyin Luo","Jie Fu","James Glass"],"abstract":"The back-propagation (BP) algorithm has been considered the de-facto method\nfor training deep neural networks. It back-propagates errors from the output\nlayer to the hidden layers in an exact manner using the transpose of the\nfeedforward weights. However, it has been argued that this is not biologically\nplausible because back-propagating error signals with the exact incoming\nweights are not considered possible in biological neural systems. In this work,\nwe propose a biologically plausible paradigm of neural architecture based on\nrelated literature in neuroscience and asymmetric BP-like methods.\nSpecifically, we propose two bidirectional learning algorithms with trainable\nfeedforward and feedback weights. The feedforward weights are used to relay\nactivations from the inputs to target outputs. The feedback weights pass the\nerror signals from the output layer to the hidden layers. Different from other\nasymmetric BP-like methods, the feedback weights are also plastic in our\nframework and are trained to approximate the forward activations. Preliminary\nresults show that our models outperform other asymmetric BP-like methods on the\nMNIST and the CIFAR-10 datasets.","url_abs":"http://arxiv.org/abs/1702.07097v4","url_pdf":"http://arxiv.org/pdf/1702.07097v4.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":"adaptive-bidirectional-backpropagation","repo_url":"https://github.com/SkTim/bdfa-torch","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"torch","reach":null},{"paper_slug":"adaptive-bidirectional-backpropagation","repo_url":"https://github.com/bigaidream-projects/bidirectional","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}