{"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/larnn-linear-attention-recurrent-neural","title":"LARNN: Linear Attention Recurrent Neural Network","arxiv_id":"1808.05578","date":"2018-08-16","proceeding":null,"authors":["Guillaume Chevalier"],"abstract":"The Linear Attention Recurrent Neural Network (LARNN) is a recurrent\nattention module derived from the Long Short-Term Memory (LSTM) cell and ideas\nfrom the consciousness Recurrent Neural Network (RNN). Yes, it LARNNs. The\nLARNN uses attention on its past cell state values for a limited window size\n$k$. The formulas are also derived from the Batch Normalized LSTM (BN-LSTM)\ncell and the Transformer Network for its Multi-Head Attention Mechanism. The\nMulti-Head Attention Mechanism is used inside the cell such that it can query\nits own $k$ past values with the attention window. This has the effect of\naugmenting the rank of the tensor with the attention mechanism, such that the\ncell can perform complex queries to question its previous inner memories, which\nshould augment the long short-term effect of the memory. With a clever trick,\nthe LARNN cell with attention can be easily used inside a loop on the cell\nstate, just like how any other Recurrent Neural Network (RNN) cell can be\nlooped linearly through time series. This is due to the fact that its state,\nwhich is looped upon throughout time steps within time series, stores the inner\nstates in a \"first in, first out\" queue which contains the $k$ most recent\nstates and on which it is easily possible to add static positional encoding\nwhen the queue is represented as a tensor. This neural architecture yields\nbetter results than the vanilla LSTM cells. It can obtain results of 91.92% for\nthe test accuracy, compared to the previously attained 91.65% using vanilla\nLSTM cells. Note that this is not to compare to other research, where up to\n93.35% is obtained, but costly using 18 LSTM cells rather than with 2 to 3\ncells as analyzed here. Finally, an interesting discovery is made, such that\nadding activation within the multi-head attention mechanism's linear layers can\nyield better results in the context researched hereto.","url_abs":"http://arxiv.org/abs/1808.05578v1","url_pdf":"http://arxiv.org/pdf/1808.05578v1.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":"larnn-linear-attention-recurrent-neural","repo_url":"https://github.com/guillaume-chevalier/Linear-Attention-Recurrent-Neural-Network","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"time-series-1","task_name":"Time Series"},{"task_slug":"time-series","task_name":"Time Series Analysis"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1808.05578","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}