Meta tags:

Headings (most frequently used words):

inception, deep, learning, architecture, contents, version, history, references, external, links, v1, v2, v3, v4, xception,

Text of the page (most frequently used words):
the (56), inception (47), and (22), google (16), 2017 (13), edit (12), deep (11), #learning (11), architecture (11), 2024 (11), was (10), displaystyle (10), with (9), times (9), computer (8), 2016 (8), 2014 (8), network (8), that (8), wikipedia (7), this (7), 2025 (7), from (7), 2022 (7), 2015 (7), for (7), vision (6), recognition (6), 2023 (6), other (6), models (6), released (6), cvpr (6), resnet (6), cnn (6), are (6), 320 (6), version (6), data (6), page (5), neural (5), also (5), 2019 (5), conference (5), convolutions (5), arxiv (5), googlenet (5), convolution (5), module (5), factorized (5), were (5), they (5), convolutional (5), history (5), contents (4), search (4), networks (4), retrieved (4), tensorflow (4), pattern (4), all (4), processing (4), alphago (4), links (4), 978 (4), isbn (4), doi (4), ieee (4), xception (4), residual (4), published (4), layers (4), more (4), found (4), team (4), size (4), normalization (4), hide (4), move (4), sidebar (4), toggle (3), view (3), using (3), license (3), artificial (3), gemini (3), master (3), github (3), research (3), 1109 (3), image (3), szegedy (3), christian (3), vanhoucke (3), vincent (3), connections (3), official (3), repo (3), kaggle (3), training (3), can (3), both (3), model (3), distribution (3), auxiliary (3), head (3), dimension (3), layer (3), these (3), top (3), batch (3), which (3), modules (3), later (3), original (3), tools (3), main (3), languages (2), table (2), code (2), contact (2), about (2), privacy (2), policy (2), text (2), under (2), terms (2), you (2), commons (2), categories (2), short (2), description (2), wikidata (2), software (2), programming (2), learn (2), computing (2), information (2), differentiable (2), general (2), generative (2), transformer (2), need (2), 2018 (2), tensor (2), intelligence (2), machine (2), versions (2), slim (2), readme (2), external (2), com (2), depthwise (2), separable (2), zhang (2), ioffe (2), sergey (2), alexander (2), aaai (2), proceedings (2), 4673 (2), arora (2), cambridge (2), new (2), into (2), deeper (2), references (2), each (2), days (2), linear (2), design (2), proposed (2), correlations (2), where (2), empirically (2), label (2), regularization (2), classification (2), predict (2), probability (2), epsilon (2), delta (2), dots (2), entry (2), than (2), removed (2), form (2), example (2), 640 (2), single (2), stacked (2), another (2), has (2), parameters (2), help (2), improves (2), used (2), paper (2), most (2), modern (2), prediction (2), softmax (2), body (2), perform (2), stem (2), three (2), after (2), solved (2), since (2), classifiers (2), 3l_ (2), aux (2), reduced (2), apache (2), lenet (2), renamed (2), family (2), release (2), appearance (2), upload (2), file (2), changes (2), read (2), article (2), log (2), create (2), account (2), donate (2), menu (2), add, topic, mobile, cookie, statement, statistics, developers, conduct, legal, safety, contacts, disclaimers, available, additional, may, apply, site, agree, registered, trademark, non, profit, organization, wikimedia, foundation, inc, use, creative, attribution, sharealike, rendered, parsoid, last, edited, september, utc, hidden, different, articles, https, org, index, php, title, inception_, deep_learning_architecture, oldid, 1310431224, portals, technology, mindspore, flux, jax, theano, scikit, keras, pytorch, libraries, spinnaker, memristor, vpu, tpu, ipu, hardware, inductive, bias, computational, theory, ricci, calculus, neuromorphic, automatic, differentiation, statistical, manifold, geometry, category, workspace, labs, pre, trained, future, summit, attention, see, antigravity, robotics, vids, notebooklm, dreambooth, veo, genie, gemma, videopoet, imagen, palm, chinchilla, 2021, lamda, xlnet, bert, nano, banana, sparrow, assistant, chatbots, unit, quantum, lab, gato, efficientnet, mobilenet, wavenet, alphagenome, alphaevolve, alphaproof, alphageometry, funsearch, alphadev, alphatensor, alphastar, alphafold, popular, culture, jie, lee, sedol, fan, hui, competitions, muzero, alphazero, zero, programs, deepmind, brain, list, openai, compute, chollet, francois, 1258, 5386, 0457, 195, 1251, kaiming, xiangyu, ren, shaoqing, sun, jian, dec, 1512, 03385, alemi, 2374, 3468, issn, 1609, v31i1, 11231, 1602, 07261, impact, shlens, jon, wojna, zbigniew, 2826, 8851, 308, 2818, rethinking, sanjeev, bhaskara, aditya, rong, tengyu, pmlr, 592, 1310, 6343, 584, 31st, international, provable, bounds, some, representations, lin, min, chen, qiang, yan, shuicheng, 1312, 4400, aston, lipton, zachary, smola, york, port, melbourne, delhi, singapore, university, press, 009, 38943, dive, multi, branch, wei, liu, yangqing, jia, sermanet, pierre, reed, scott, anguelov, dragomir, erhan, dumitru, rabinovich, andrew, june, going, 6964, 7298594, 1409, 4842, took, k80, gpus, approximately, petaflop, extreme, stack, hypothesis, cross, channels, spatial, feature, maps, entirely, decoupled, modifications, added, inspired, incremental, update, even, complications, improve, benchmarks, smoothing, instead, making, made, smoothed, total, number, classes, underbrace, lowest, classifier, during, worked, uses, reduction, concatenating, output, downscaled, stride, maxpooling, pool, then, concatenated, pooling, factored, receptive, field, kernel, compared, just, thus, strictly, powerful, however, power, not, necessarily, needed, adding, removing, dropout, became, unnecessary, when, local, response, famous, proposing, had, million, structure, architectures, final, fully, connected, produces, next, many, bulk, ingestion, first, few, preprocessing, downscale, images, smaller, consists, parts, one, complete, suffered, two, inserted, within, loss, function, weighted, sum, real, vanishing, gradient, problem, composed, stated, logical, culmination, individual, left, standard, right, name, came, 1998, cnns, called, internet, meme, phrase, film, because, again, 2010, developed, instance, won, ilsvrc14, imagenet, large, scale, visual, challenge, introduced, researchers, series, historically, important, early, separates, ingest, architectural, persists, blob, repository, type, stable, author, free, encyclopedia, item, projects, printable, download, pdf, print, export, switch, legacy, parser, get, shortened, url, cite, permanent, link, related, what, here, actions, english, talk, bahasa, indonesia, català, subsection, personal, special, pages, recent, community, portal, contribute, random, current, events, navigation, jump, content,

Text of the page (random words):
inception deep learning architecture wikipedia jump to content main menu main menu move to sidebar hide navigation main page contents current events random article about wikipedia contact us contribute help learn to edit community portal recent changes upload file special pages search search appearance donate create account log in personal tools donate create account log in contents move to sidebar hide top 1 version history toggle version history subsection 1 1 inception v1 1 2 inception v2 1 3 inception v3 1 4 inception v4 1 5 xception 2 references 3 external links toggle the table of contents inception deep learning architecture 3 languages català bahasa indonesia 中文 edit links article talk english read edit view history tools tools move to sidebar hide actions read edit view history general what links here related changes upload file permanent link page information cite this page get shortened url switch to legacy parser print export download as pdf printable version in other projects wikidata item appearance move to sidebar hide from wikipedia the free encyclopedia family of convolutional neural networks inception original author google ai release 2014 stable release v4 2017 type convolutional neural network license apache 2 0 repository github com tensorflow models blob master research slim readme md inception 1 is a family of convolutional neural network cnn for computer vision introduced by researchers at google in 2014 as googlenet later renamed inception v1 the series was historically important as an early cnn that separates the stem data ingest body data processing and head prediction an architectural design that persists in all modern cnn 2 inception v3 model version history edit inception v1 edit googlenet architecture in 2014 a team at google developed the googlenet architecture an instance of which won the imagenet large scale visual recognition challenge 2014 ilsvrc14 1 3 the name came from the lenet of 1998 since both lenet and googlenet are cnns they also called it inception after a we need to go deeper internet meme a phrase from inception 2010 the film 1 because later more versions were released the original inception architecture was renamed again as inception v1 the models and the code were released under apache 2 0 license on github 4 an individual inception module on the left is a standard module and on the right is a dimension reduced module a single inception dimension reduced module the inception v1 architecture is a deep cnn composed of 22 layers most of these layers were inception modules the original paper stated that inception modules are a logical culmination of network in network 5 and arora et al 2014 6 since inception v1 is deep it suffered from the vanishing gradient problem the team solved it by using two auxiliary classifiers which are linear softmax classifiers inserted at 1 3 deep and 2 3 deep within the network and the loss function is a weighted sum of all three l 0 3 l a u x 1 0 3 l a u x 2 l r e a l displaystyle l 0 3l_ aux 1 0 3l_ aux 2 l_ real these were removed after training was complete this was later solved by the resnet architecture the architecture consists of three parts stacked on top of one another 2 the stem data ingestion the first few convolutional layers perform data preprocessing to downscale images to a smaller size the body data processing the next many inception modules perform the bulk of data processing the head prediction the final fully connected layer and softmax produces a probability distribution for image classification this structure is used in most modern cnn architectures inception v2 edit inception v2 was released in 2015 in a paper that is more famous for proposing batch normalization 7 8 it had 13 6 million parameters it improves on inception v1 by adding batch normalization and removing dropout and local response normalization which they found became unnecessary when batch normalization is used inception v3 edit inception v3 was released in 2016 7 9 it improves on inception v2 by using factorized convolutions as an example a single 5 5 convolution can be factored into 3 3 stacked on top of another 3 3 both has a receptive field of size 5 5 the 5 5 convolution kernel has 25 parameters compared to just 18 in the factorized version thus the 5 5 convolution is strictly more powerful than the factorized version however this power is not necessarily needed empirically the research team found that factorized convolutions help it also uses a form of dimension reduction by concatenating the output from a convolutional layer and a pooling layer as an example a tensor of size 35 35 320 displaystyle 35 times 35 times 320 can be downscaled by a convolution with stride 2 to 17 17 320 displaystyle 17 times 17 times 320 and by maxpooling with pool size 2 2 displaystyle 2 times 2 to 17 17 320 displaystyle 17 times 17 times 320 these are then concatenated to 17 17 640 displaystyle 17 times 17 times 640 other than this it also removed the lowest auxiliary classifier during training they found that the auxiliary head worked as a form of regularization they also proposed label smoothing regularization in classification for an image with label c displaystyle c instead of making the model to predict the probability distribution δ c 0 0 0 1 c th entry 0 0 displaystyle delta _ c 0 0 dots 0 underbrace 1 _ c text th entry 0 dots 0 they made the model predict the smoothed distribution 1 ϵ δ c ϵ k displaystyle 1 epsilon delta _ c epsilon k where k displaystyle k is the total number of classes inception v4 edit in 2017 the team released inception v4 inception resnet v1 and inception resnet v2 10 inception v4 is an incremental update with even more factorized convolutions and other complications that were empirically found to improve benchmarks inception resnet v1 and v2 are both modifications of inception v4 where residual connections are added to each inception module inspired by the resnet architecture 11 xception edit xception extreme inception was published in 2017 12 it is a linear stack of depthwise separable convolution layers with residual connections the design was proposed on the hypothesis that in a cnn the cross channels correlations and spatial correlations in the feature maps can be entirely decoupled training each network took 3 days on 60 k80 gpus or approximately 0 5 petaflop days 13 references edit 1 2 3 szegedy christian wei liu yangqing jia sermanet pierre reed scott anguelov dragomir erhan dumitru vanhoucke vincent rabinovich andrew june 2015 going deeper with convolutions 2015 ieee conference on computer vision and pattern recognition cvpr ieee pp 1 9 arxiv 1409 4842 doi 10 1109 cvpr 2015 7298594 isbn 978 1 4673 6964 0 1 2 zhang aston lipton zachary li mu smola alexander j 2024 8 4 multi branch networks googlenet dive into deep learning cambridge new york port melbourne new delhi singapore cambridge university press isbn 978 1 009 38943 3 official repo of inception v1 on kaggle published by google google inception google 2024 08 19 retrieved 2024 08 19 lin min chen qiang yan shuicheng 2014 03 04 network in network arxiv 1312 4400 cs ne arora sanjeev bhaskara aditya ge rong ma tengyu 2014 01 27 provable bounds for learning some deep representations proceedings of the 31st international conference on machine learning pmlr 584 592 arxiv 1310 6343 1 2 szegedy christian vanhoucke vincent ioffe sergey shlens jon wojna zbigniew 2016 rethinking the inception architecture for computer vision 2016 ieee conference on computer vision and pattern recognition cvpr pp 2818 2826 doi 10 1109 cvpr 2016 308 isbn 978 1 4673 8851 1 official repo of inception v2 on kaggle published by google official repo of inception v3 on kaggle published by google szegedy christian ioffe sergey vanhoucke vincent alemi alexander 2017 02 12 inception v4 inception resnet and the impact of residual connections on learning proceedings of the aaai conference on artificial intelligence 31 1 arxiv 1602 07261 doi 10 1609 aaai v31i1 11231 issn 2374 3468 he kaiming zhang xiangyu ren shaoqing sun jian 10 dec 2015 deep residual learning for image recognition arxiv 1512 03385 cs cv chollet francois 2017 xception deep learning with depthwise separable convolutions 2017 ieee conference on computer vision and pattern recognition cvpr pp 1251 1258 doi 10 1109 cvpr 2017 195 isbn 978 1 5386 0457 1 ai and compute openai com 2022 06 09 retrieved 2025 04 28 external links edit a list of all inception models released by google models research slim readme md at master tensorflow models github retrieved 2024 10 19 v t e google ai google google brain google deepmind computer programs alphago versions alphago 2015 master 2016 alphago zero 2017 alphazero 2017 muzero 2019 competitions fan hui 2015 lee sedol 2016 ke jie 2017 in popular culture alphago 2017 other alphafold 2018 alphastar 2019 alphatensor 2022 alphadev 2023 funsearch 2023 alphageometry 2024 alphaproof 2024 alphaevolve 2025 alphagenome 2025 machine learning neural networks inception 2014 wavenet 2016 mobilenet 2017 transformer 2017 efficientnet 2019 gato 2022 other quantum artificial intelligence lab tensorflow tensor processing unit generative ai chatbots assistant 2016 sparrow 2022 gemini 2023 nano banana 2025 models bert 2018 xlnet 2019 t5 2019 lamda 2021 chinchilla 2022 palm 2022 imagen 2023 gemini 2023 videopoet 2024 gemma 2024 genie 2024 veo 2024 other dreambooth 2022 notebooklm 2023 vids 2024 gemini robotics 2025 antigravity 2025 see also attention is all you need future of go summit generative pre trained transformer google labs google workspace category commons v t e differentiable computing general differentiable programming information geometry statistical manifold automatic differentiation neuromorphic computing pattern recognition ricci calculus computational learning theory inductive bias hardware ipu tpu vpu memristor spinnaker software libraries tensorflow pytorch keras scikit learn theano jax flux jl mindspore portals computer programming technology retrieved from https en wikipedia org w index php title inception_ deep_learning_architecture oldid 1310431224 categories artificial neural networks computer vision google software hidden categories articles with short description short description is different from wikidata this page was last edited on 9 september 2025 at 15 12 utc page was rendered with parsoid text is available under the creative commons attribution sharealike 4 0 license additional terms may apply by using this site you agree to the terms of use and privacy policy wikipedia is a registered trademark of the wikimedia foundation inc a non profit organization privacy policy about wikipedia disclaimers contact wikipedia legal safety contacts code of conduct developers statistics cookie statement mobile view search search toggle the table of contents inception deep learning architecture 3 languages add topic