LINE Data Labs - Possibility of AI Technology

LINE Corporation：https://linecorp.com/ja/
LINE Engineering：https://engineering.linecorp.com/ja/
LINE 採用情報：https://linecorp.com/ja/career/

Transcript

1. LINE Data Labs Possibility of AI Technology ೔ຊޠ
2. ࢲͨͪͷϛογϣϯ͸ɺੈքதͷਓͱਓɺਓͱ৘ใɾαʔϏεͱͷڑ཭ΛॖΊΔ͜ͱͰ͢ɻ LINE͸ɺ 2011೥3݄11೔ʹ೔ຊͰൃੜͨ͠਒ࡂΛ͖͔͚ͬʹɺ ϞόΠϧϝοηϯδϟʔͱͯ͠ੜ·Ε·ͨ͠Ň Ո଒΍༑ਓɾ࿀ਓͳͲɺ਎ۙʹ͍Δେ੾ͳਓͱͷؔ܎ੑΛਂΊɺឺΛڧ͘͢Δίϛϡχέʔγϣϯखஈ ͕ͦ͜ɺ೔ຊͷΈͳΒͣੈքதͰٻΊΒΕ͍ͯΔͱߟ͑ɺ2011೥6݄ʹαʔϏεΛ։࢝͠·ͨ͠ɻ͍·΍ɺ LINE͸೔ຊΛ͸͡Ίɺ ΞδΞ஍ҬΛத৺ʹྗڧ͍੒௕Λଓ͚ɺ ੈքͰ਺ԯਓʹར༻͞ΕΔϞόΠϧϝοηϯ δϟʔ΁ͱ੒௕͍ͯ͠·͢ɻ ࢲͨͪ͸ɺ ୯ʹੈքڞ௨ʹըҰԽ͞ΕͨάϩʔόϧαʔϏεͰ͸ͳ͘ɺ ֤ࠃͷଟछଟ༷ͳจԽɾ׳शΛଚॏ͠Ŋ lαʔϏεΛΧϧνϟϥΠζ ʢจԽԽʣ z͍ͯ͘͜͠ͱ͕ͦ͜ɺ ͦͷࠃͰɺ ͦͷ౔஍ͰɺαʔϏεΛਅʹࠜ෇͔ͤŊ ੒௕͍ͤͯͨ͘͞Ίʹେ੾ͳ͜ͱͩͱڧ͘৴͍ͯ͡·͢ɻ͜Ε͔ΒɺϞόΠϧϝοηϯδϟʔͱͯ͠஀ੜͨ͠ LINE͸ɺϞόΠϧ্ͷϢʔβʔχʔζΛ౷߹తʹղܾ͍ͯ͘͠ɺεϚʔτϙʔλϧ΁ͱਐԽ͍͖ͯ͠·͢ɻ ਓ΍৘ใɾίϯςϯπɺΦϯϥΠϯɾΦϑϥΠϯαʔϏεͳͲ͋ΒΏΔ΋ͷͱϢʔβʔ͕͍ͭͰ΋ɺͲ͜Ͱ΋Ŋ ࠷దͳڑ཭ͰγʔϜϨεʹܨ͕͍ͬͯ͘ɺ LINEΛೖΓޱͱͯ͠ੜ׆ͷશ͕ͯ׬݁͢Δੈքͷ࣮ݱΛ໨ࢦͯ͠ ͍͖·͢ɻ LINE͸ɺ͋ͳͨͱੈքΛͭͳ͗ɺຖ೔Λڞʹաͯ͘͝͠ΕΔɺ ͦΜͳɺՈ଒΍༑ਓͷΑ͏ͳଘࡏʹͳΓ͍ͨͱ ࢥ͍ͬͯ·͢ɻ About LINE
3. LINEͰ͸ɺ ϝοηϯδϟʔΛؚΊͨશαʔϏεͷσʔλ෼ੳɾݚڀɾԠ༻ͷͨΊʹઐ໳తͳ ૊৫Λฤ੒͍ͯ͠·͢ɻ ֤αʔϏεͷϩά΍ར༻ಈ޲ͳͲͷσʔλΛࣄۀྖҬΛ௒͑ͯ ԣஅతʹॲཧ͠ɺΑΓਫ਼ີͳσʔλ෼ੳͱ৘ใϑΟϧλϦϯάΛఏڙ͢Δ͜ͱͰɺશαʔ Ϗεͷσʔλͷޮ཰తͳ׆༻Λ໨ࢦ͍ͯ͠·͢ɻ σʔλϓϥοτϑΥʔϜͷઃܭɾ։ൃɾӡӦΛ͸͡Ίͱͯ͠ɺ஝ੵ͞Εͨσʔλͷਫ਼੡Ŋ ػցֶशΛج൫ʹͨ͠ίϯςϯπਪનɾϑΟϧλϦϯάɾϢʔβʔλʔήςΟϯάͳͲͷ ߴ౓ͳσʔλ෼ੳΛߦ͍ɺࣄۀઓུͷߏங΍ࣄۀͷ࠷దԽ࡞ۀΛ֤ࣄۀ૊৫ͱͱ΋ʹ ਱ߦ͍ͯ͠·͢ɻ Data ScienceνʔϜͰ͸ɺLINEͷαʔϏεՁ஋ͷ࠷େԽΛσʔλ෼ੳʹΑ࣮ͬͯݱͯ͠ ͍·͢ɻLINEͷ֤αʔϏε΍νʔϜԣஅͷϓϩδΣΫτʹ͓͍ͯɺॏཁͳҙࢥܾఆͷࢧԉŊ ՝୊ͷղܾࡦͷఏҊͱ࣮ߦɺࢪࡦͷݕূͱվળΛҰؾ௨؏Ͱ࣮ࢪ͢Δ͜ͱͰɺΑΓྑ͍ αʔϏεͮ͘ΓΛՃ଎ͤ͞Δ͜ͱ͕ϛογϣϯͰ͢ɻ LINEͷϝοηϯδϟʔ͸ඇৗʹ େن໛ͳϢʔβʔج൫Λ͓࣋ͬͯΓɺ ͜ΕʹՃ͑ͯɺଟ਺ͷϑΝϛϦʔαʔϏε͔Βඇৗʹ ଟ༷ͳσʔλ͕ಘΒΕ·͢ɻ͜ͷੈքతن໛ͷϢʔβʔσʔλͷ෼ੳΛ௨ͯ͠LINEͷ ࣄۀͷൃలʹߩݙͰ͖Δ͜ͱ͕ɺେ͖ͳັྗͷҰͭͰ͢ɻ ҰํɺMachine LearningνʔϜͰ͸ɺطଘαʔϏεͷվળ΍৽نαʔϏεͷاըΛ ߦ͏ͨΊͷɺֵ৽తͳػցֶश͓ΑͼσʔλϚΠχϯάٕज़Λ։ൃ͍ͯ͠·͢ɻػց ֶशʹؔ͢Δ৽͍ٕ͠ज़ɾ஌ݟΛ׆͔͠ɺαʔϏεվળ΍৽ͨͳاըͷ࣮ݱΛݕ౼ɾ࣮ફ ͍ͯ͠·͢ɻ Data Science / Machine Learning
4. Research LINEͰ͸ɺࠓޙࣄۀΛ֦େ͢Δ্ͰجૅͱͳΔٕज़ɺಛʹਓ޻஌ೳʹؔΘΔݚڀΛਐΊ͍ͯ·͢ɻ۩ମ తͳݚڀྖҬ͸ɺػցֶशΛ࣠ʹɺԻ੠ॲཧɺݴޠॲཧɺը૾ॲཧͳͲͰ͢ɻ Ұൠతʹɺاۀ಺ݚڀॴ͸ݚڀ੒Ռͷ੡඼ԽΛ໨తͱ͍ͯ͠Δ͜ͱ͕ଟ͍Ͱ͕͢ɺ LINEͰ͸ɺ੡඼Խʹ ௚݁͢ΔݚڀͰ͸ͳ͘ɺٕज़։ൃ ɾൃ໌Λ໨తͱ͍ͯ͠·͢ɻ ݚڀһ͸·ͩগ਺Ͱ͕͢ɺݚڀςʔϚʹ ͍ͭͯ͸ɺ೔ຊࠃ಺ͷଟ਺ͷେֶͱ࿈ܞ͠ɺڞಉݚڀΛߦ͏ମ੍Λඋ͍͑ͯ·͢ɻࠓޙ͸ւ֎ͷݚڀػؔ ͱͷ࿈ܞ΋ࢹ໺ʹೖΕ͍ͯ·͢ɻ ݚڀһ͸ʮࠃࡍձٞͰ࿦จ͕࠾୒͞ΕͯධՁ͞ΕΔ͜ͱʯ͕།ҰͷϛογϣϯͰ͢ɻձ͔ࣾΒͷධՁͱŊ ݚڀһݸਓͱͯ͠ͷ࣮੷͕࿈ಈͨ͠ΫϦΞͳධՁମܥ͕ͻ͔Ε͓ͯΓɺࣥචʹ຅಄Ͱ͖ΔΑ͏ɺΦϑΟε͸ ϦϥοΫεͰ͖Δ؀ڥ͕༻ҙ͞Ε͍ͯ·͢ɻݚڀҎ֎ͷࡶ຿΍ଧͪ߹Θͤʹ೰·͞ΕΔ͜ͱ͸͋Γ·ͤΜŇ ͱ͸͍͑ৗʹࣾ಺ʹด͜͡΋Δ͚ͩͰ͸ͳ͘ɺڞಉݚڀͷͨΊʹ͍ΖΜͳେֶΛ๚໰͠ɺ࠷ઌ୺ͷݚڀʹ ৮ΕΔػձ΋ଟ͍Ͱ͢ɻ·ͨLINEͰ͸ɺࣾ಺֎͔ΒAIྖҬͷୈҰਓऀΛট଴͢ΔΠϕϯτͳͲ΋සൟʹ ։࠵͓ͯ͠Γɺ֎෦͔Βͷܹࢗ͸ଟ͍؀ڥͰ͢ɻ اۀʹ͍ͳ͕ΒΞΧσϛοΫͳݚڀʹܞΘΓ͍ͨํʹ࠷దͳ؀ڥͰ͢ɻݚڀҙཉ͕Ԣ੝Ͱܹࢗతͳ؀ڥʹ ඈͼࠐΜͰΈ͍ͨํ͸ɺͥͻ͝Ԡื͍ͩ͘͞ʂ LINE' Selected Publications ICASSP2020 1 M. Togami, Y. Masuyama, T. Komatsu, and Y. Nakagome, "Unsupervised Training for Deep Speech Source Separation with Kullback-Leibler Divergence based Probabilistic Loss Function" 2 M. Togami, "Multi-channel Speech Source Separation and Dereverberation with Sequential Integration of Determined and Underdetermined Models" 3 Y. Masuyama, M. Togami, and T. Komatsu "Consistency-Aware Multi-channel Speech Enhancement using Deep Neural Networks" 4 Y. Nakagome, M. Togami, T. Ogawa, T. Kobayashi, "Deep Speech Extranction with Time-varying Spatial Filtering Guided by Desired Direction Attractor" 5 T. Komatsu, K. Imoto, M. Togami, "Scene-dependent Acoustic Event Detection with Scene Conditioning and Fake- scene-conditioned Loss" 6 K. Miyazaki, T. Komatsu, T. Hayashi, S. Watanabe, T. Toda, K. Takeda, "Weakly-Supervised Sound Event Detection with Self-attention" 7 R. Yamamoto, E. Song, J.-M. Kim, "Parallel WaveGAN: A fast waveform generation model based on generative adversarial networks with multi-resolution spectrogram" 8 M.-J. Hwang, E. Song, R. Yamamoto, F. Soong, H.-G. Kang, "Improving LPCNet-based Text-to-Speech with Linear Prediction-structured Mixture Density Network" BigData2019 9 T. Takahashi, "Indirect Adversarial Attacks via Poisoning Neighbors for Graph Convolutional Networks" INTERSPEECH2019 10 Y. Masuyama, M. Togami, and T. Komatsu, "Multichannel Loss Function for Supervised Speech Source Separation by Mask-based Beamforming" 11 M. Togami and T. Komatsu, "Variational Bayesian Multi-channel Speech Dereverberation under Noisy Environments with Probabilistic Convolutive Transfer Function" 12 R. Yamamoto, E. Song, and J-.M. Kim, "Probability Density Distillation with Generative Adversarial Networks for High- Quality Parallel Waveform Generation" SIGIR2019 13 W. Sakata, T. Shibata, R. Tanaka, and S. Kurohashi, "FAQ Retrieval using Query-Question Similarity and BERT-Based Query-Answer Relevance" ICASSP2019 14 M. Togami, “Spatial Constraint on Multichannel Deep Clustering" 15 M. Togami, “Multichannel Itakura Saito Distance Minimization with deep neural network”
5. ňClovaʯ(Ϋϩʔό ʣ͸ɺ೔ຊΛ͸͡ΊɺΞδΞͰτοϓγΣΞΛ࣋ͭίϛϡχέʔγϣϯΞϓϦͰ͋Δ ňLINEʯ ͱؖࠃNo.1ͷϙʔλϧαΠτ ʮNAVERʯ ͕ڞಉͰ։ൃ͍ͯ͠ΔAIΞγελϯτͰ͢ɻLINEͱNAVER͕ ௕೥ʹΘͨΓݚڀΛଓ͚͍ͯΔੈք࠷ߴϨϕϧͷݕࡧɺԻ੠ೝࣝɾԻ੠߹੒ɺը૾ɾಈըͷ෼ੳɺࣗવݴޠ ॲཧ͓Αͼձ࿩ͳͲͷAIٕज़ʹՃ͑ɺLINEͱNAVER͕อ༗͍ͯ͠Δ਺ଟ͘ͷίϯςϯπ΍αʔϏεΛ׆༻ ͢Δ͜ͱʹΑΓɺ ΑΓεϚʔτͳAIΞγελϯτΛ࣮ݱ͠·ͨ͠ɻ ·ͨɺ ୭΋͕ؾܰʹClovaϓϥοτϑΥʔϜΛ ׆༻ͯ͠৽ͨͳػೳʢClovaεΩϧʣ Λ࡞ΕΔΑ͏ʹ͓ͯ͠ΓɺϢʔβʔ͓Αͼ༷ʑͳύʔτφʔͱͱ΋ʹŊ ͞ΒʹศརͳੈքΛ࡞Ϧ্͛Δ͜ͱʹऔΓ૊ΜͰ͍·͢ɻ LINE͸ɺ͜Ε·ͰʹAIΞγελϯτʮClovaʯΛ͸͡Ίͱͯ͠ɺίϛϡχέʔγϣϯΞϓϦʮLINEʯ΍֤ αʔϏεʹ͓͍ͯਂ૚ֶशΛϕʔεͱ༷ͨ͠ʑͳAIٕज़ͷ։ൃɺ౥ࡌΛਐΊ͖ͯ·ͨ͠ɻ ʮLINE BRAIN ʼn ࣄۀͰ͸ɺ͜Ε·ͰLINE͕ഓ͖ͬͯͨAIٕज़Λ֎෦اۀ౳ʹ޲͚ͯఏڙ͍ͯ͠·͢ɻ LINE BRAIN CHATBOT LINEެࣜΞΧ΢ϯτ౳ͱ࿈ܞͯ͠ར༻Ͱ͖ΔʮࣗવݴޠΛཧղ͠Ԡ౴͢Δٕज़ ʯͰ͢ɻର࿩ܗࣜͰ໨తୡ੒Λαϙʔτ ͠·͢ɻӡ༻ͷதͰAIֶ͕श͠ɺΑΓద੾ͳԠ౴Λ࣮ݱͰ͖ΔΑ͏ʹͳΓ·͢ɻ ʮ LINE BRAIN CHATBOTʯ͸ɺଞࣾͷ νϟοτϘοτٕज़ͱൺֱͨ͠ࡍʹ΋ɺ࣭໰ʹର͢Δճ౴ʹ͓͍ͯߴ͍ਖ਼ղ཰Λތ͍ͬͯ·͢ɻ LINE BRAIN OCR ࠃࡍձٞͰੈք࠷ߴਫ४ͷೝࣝਫ਼౓ΛތΔʮ ը૾಺ͷจࣈ΍จষΛςΩετσʔλ΁ม׵͢Δٕज़ ʯͰ͢ɻࣸਅ΍ εΩϟϯͨ͠ҹ࡮จࣈɺεΫϦʔϯγϣοτ಺ͷจࣈΛਫ਼౓ߴ͘ೝࣝ͠·͢ɻʮLINE BRAIN OCRʯ͸ɺICDARͰAI ڝ߹ଞࣾΛ཈͑4෼໺ͰੈքNo.1Λ֫ಘ͍ͯ͠·͢ ʢ2019. 03. 29࣌఺ ɻ LINE BRAIN SPEECH TO TEXT ి࿩΍VoIPܦ༝ͰͷԻ੠௨࿩ʹ΋ରԠͨ͠ʮਓͷൃ࿩Λจࣈʹม׵͢Δٕज़ʯͰ͢ɻίϯλΫτηϯλʔʹ͓͚ΔϦΞϧ λΠϜͷԻ੠ࣗಈԠ౴΍ɺόονॲཧͰͷ௕จॻ͖ى͜͠ʹ΋ରԠͰ͖·͢ɻ੡඼໊ͳͲݻ༗ͳൃ࿩ͷೝࣝਫ਼౓վળ χʔζʹ΋Ԡ͡·͢ɻ Clova LINE BRAIN
6. Benefit and Support Programs ࣾһҰਓͻͱΓ͕৺਎ͱ΋ʹ݈߁ͳঢ়ଶͰ࢓ࣄʹूதͰ͖ΔΑ͏ɺ අ༻ิॿ΍ػձఏڙͰαϙʔτ͍ͯ͠·͢ɻ·ͨΑΓԁ׈ʹಇͨ͘Ίʹɺ ͞·͟·ͳ੍౓Λઃ͚͍ͯ·͢ɻ ࣾ಺਌ກඅ ࣾһಉ࢜ͷίϛϡχέʔγϣϯ׆ੑԽͷͨΊɺ ׻ૹܴձ΍ϥϯν౳ͷඅ༻ΛҰ෦ิॿ͢Δ੍౓͕͋Γ·͢Ň ϦϑϨογϡٳՋ ೖ͔ࣾΒຬ5೥͝ͱʹɺಛผٳՋΛ෇༩͓ΑͼϦϑϨογϡิॿۚΛࢧڅ͠·͢ɻҰఆͷظؒ೔ৗۀ຿ ͔Β཭Εɺ৺਎ͷϦϑϨογϡͳͲࣗ༝ͳ࣌ؒΛա͍͚ͯͨͩ͝͠·͢ɻ ࣾ಺ެื ೖ͔ࣾΒҰఆظؒΛܦաͨࣾ͠һΛର৅ʹɺ֘౰ظؒʹࣾ಺Ͱެ։͞ΕΔืूϙδγϣϯʹԠืͯ͠ બߟʹ௨Ε͹ɺ ࣗ෼ͷҙࢤͰҟಈͰ͖Δ੍౓͕͋Γ·͢ɻͳ͓ɺ બߟաఔ΍ͦͷ߹൱൑அʹݱॴଐ্௕͸ ؔ༩͠·ͤΜɻ ֤छαʔϏεར༻ࢧԉ ࣾһ͕֤छαʔϏεΛΑΓؾܰʹར༻Ͱ͖ΔΑ͏ʹɺ LINE GAME΍LINEϚϯΨɺLINEελϯϓͳͲͷ ༗ྉίϯςϯπ͕ߪೖͰ͖ΔʮLINE STOREʯͰར༻Ͱ͖ΔΪϑτίʔυʢ೥ֹ24,000ԁ෼ʣΛࢧԉ ͍ͯ͠·͢ɻ LINEϞόΠϧճઢඅࢧԉ LINEϞόΠϧΛܖ໿͍ͯ͠Δࣾһʹରͯ͠ɺҰਓ1ճઢ·Ͱͷֹ݄جຊར༻ྉۚΛձ͕ࣾෛ୲͠·͢ɻ ޠֶྗΞοϓࢧԉ ࣾ಺Ͱबۀ࣌ؒલޙʹӳޠɾؖࠃޠɾ೔ຊޠͷޠֶߨ࠲Λ։ߨ͓ͯ͠Γɺશֹձࣾෛ୲ʢ5ׂҎ্ͷ ೔ఔΛग़੮ɾडߨ͕৚݅ʣͰडߨ͕ՄೳͰ͢ɻ·ͨɺTOEICͷࣾ಺ࢼݧͷػձΛઃ͚͓ͯΓɺडݧྉ͸ શֹձࣾෛ୲Ͱ͢ɻ
7. Working Environment ࣾһ͕݈߁͔ͭշదʹಇ͚ΔͨΊͷࣾ಺αʔϏεɾઃඋɾۀ຿αϙʔτ ಺༰Λ͝঺հ͠·͢ɻ ˞ Ұ෦ΦϑΟεͰ͸ɺରԠ͍ͯ͠ͳ͍΋ͷ΋͋Γ·͢ ࣾ಺ΧϑΣ ࣾ಺ʹٳܜ ɾަྲྀͳͲ͕Ͱ͖ΔΧϑΣεϖʔε͕͋Γɺ௿Ձ֨ͰίʔώʔͳͲҿΉ͜ͱ͕Ͱ͖·͢Ň ؆୯ͳձٞ΍࡞ۀ؀ڥͱͯ͠΋ར༻Ͱ͖·͢ɻ·͓ͨʹ͗Γ΍ύϯͳͲΛຖேແྉͰఏڙ͍ͯ͠·͢Ň ঢ߱ࣜσεΫ ϫϯλονͰߴ͕͞ม͑ΒΕΔσεΫͰ͢ɻனؒ͸࠲ͬͯɺ༦ํ͔Β͸ؾ෼స׵Ͱཱͬͯ࢓ࣄͳͲŊ ࣗ༝ͳελΠϧͰಇ͘͜ͱ͕Ͱ͖·͢ɻΞʔϩϯνΣΞΛ͸͡Ίͱ͢ΔߴػೳΦϑΟενΣΞͰ࢓ࣄΛ ͢Δ͜ͱ͕Ͱ͖·͢ɻ ւ֎ࢧԉ ඈߦػ΍ϗςϧͷख഑ͳͲͷग़ுαϙʔτɺॅډܖ໿΍ۓٸ࣌ͷಈ޲ͳͲͷ֎ࠃ੶ࣾһͷੜ׆αϙʔτŊ Ո଒ؚΉ৽نऔಘ΍มߋͳͲͷVISAαϙʔτΛߦ͍·͢ɻ ௨༁ɾ຋༁ ௨༁Λඞཁͱ͢Δଧͪ߹Θͤ΍໘઀΁ͷಉ੮΍ɺࣾ಺֎ͷެࣜจॻͷ຋༁ͳͲɺϓϩͷ௨༁ɾ຋༁ εΩϧΛ࣋ͭࣾһ͕ࡏ੶͠ɺ༷ʑͳίϛϡχέʔγϣϯΛαϙʔτ͍ͯ͠·͢ɻ LINE CARE έʔϒϧͳͲͷITඋ඼΍จ๪۩ͳͲͷඋ඼Ϩϯλϧɺࣾ಺Ͱͷམͱ͠෺΍खଓ͖ํ๏ͷ૬ஊͳͲۀ຿ ্Ͱࠔͬͨ࣌ʹɺLINEͷνϟοτ΍ରਓΧ΢ϯλʔͳͲͷ૬ஊ૭ޱͰۀ຿Λαϙʔτ͍ͯ͠·͢ɻ
8. LINE creates WOW to the world LINE Corporation https://linecorp.com/ja/ LINE Engineering https://engineering.linecorp.com/ja/ LINE ࠾༻৘ใ https://linecorp.com/ja/career/