You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
=> "Spatial: A Language and Compiler for Application Accelerators", PLDI, Jun 21, 2018 https://ppl.stanford.edu/papers/spatial18.pdf … Slides https://ppl.stanford.edu/papers/pldi_18_spatial_slides.pdf … https://www.youtube.com/watch?v=A94ChZ8Kv34 … https://github.com/stanford-ppl/spatial-lang … Designing Computer Sys for SW 2.0, K. Olukotun, Keynote, ISCA 2018 https://twitter.com/ogawa_tter/status/1007273981307596800 …pic.twitter.com/MmIbSgDowU
=> "LeFlow: Enabling Flexible FPGA High-Level Synthesis of Tensorflow Deep Neural Networks", arXiv, Jul 14, 2018 https://arxiv.org/abs/1807.05317 https://github.com/danielholanda/LeFlow … https://www.youtube.com/watch?v=y76_RIyq4TM … Int WS on FPGAs for Software Programmers (FSP 2018), Aug 31, 2018 http://www.fsp-workshop.org/ pic.twitter.com/Hxx1UFONA8
=> "DeepCPU: Serving RNN-based Deep Learning Models 10x Faster", Microsoft AI and Research, USENIX ATC 2018, Jul 13, 2018 https://www.usenix.org/system/files/conference/atc18/atc18-zhang-minjia.pdf … Slides https://www.usenix.org/sites/default/files/conference/protected-files/atc18_slides_zhang_0.pdf … DeepCPU improves latency and efficiency It also empowers CPUs to beat GPUs on RNN serving Open Source?pic.twitter.com/db5C36WGR6
=> "Accelerating the Exploration of Promising Artificial Intelligence Concepts New effort will expedite pioneering AI research, rapidly moving from idea to award", DAPRA, Jul 20, 2018 https://www.darpa.mil/news-events/2018-07-20a … AI Exploration will constitute a series of unique funding opportunities
=> 「AIチップ開発加速のためのイノベーション推進事業／【研究開発項目〔1〕】AIチップに関するアイディア実用化に向けた開発」に係る公募、NEDO、平成30年7月23日 http://www.nedo.go.jp/koubo/CA2_100187.html … 平成30年7月23日～9月3日 説明会資料 http://www.nedo.go.jp/content/100881313.pdf … 5000万円以内×2年間 ①～⑤すべてを満たす必要pic.twitter.com/1aLOJNbE8l
=> "Machine Learning on Google Cloud", Google, CERN Computing Seminar, Jul 17, 2018 PDF https://indico.cern.ch/event/740298/attachments/1692270/2722990/Final_Machine_Learning_on_Google_Cloud_.pdf … 80 pages (20 MB) 15+ Yrs of Solving Data Problems End to End ML Pipeline Google ML系サービスの全体像！ M. Warrick, Co-Hosts of GCP Podcast https://gcppodcast.com/ pic.twitter.com/wDbDvfG997
=> Google Edge TPU, Jul 25, 2018 https://twitter.com/ogawa_tter/status/1022468275937005568 … Edge TPU Dev Board: SOM (40 x 48 mm) + Baseboard (85 x 56 mm) Edge TPU Accelerator:65 mm x 30 mm Connector: USB Type-C (data/power)
=>
Prof Steve Furber, Designer of the Original ARM Microprocessor, Joins @Tachyum Board of Advisors, Jul 26, 2018 http://www.tachyum.com/pr-2018-07-26.shtml …
Oral History https://twitter.com/ogawa_tter/status/789027931825725441 …
Video https://twitter.com/ogawa_tter/status/856780386185039872 …
SpiNNaker https://twitter.com/ogawa_tter/status/1018859985030139909 …
Tachyum, May 2018https://twitter.com/ogawa_tter/status/999668393060909056 …
=> "In-Package Domain-Specific ASICs for Intel Stratix 10 FPGAs: A Case Study of Accelerating Deep Learning Using TensorTile ASIC", Intel, Full Paper, FPL 2018, Aug 27, 2018 Poster, FPGA 2018https://twitter.com/ogawa_tter/status/993337328419586054 …
=> "HLS4ML: deploying deep learning on FPGAs for L1 trigger and Data Acquisition", CERN, EP-IT Data science seminars, Apr 25, 2018 PDF https://indico.cern.ch/event/721567/attachments/1631390/2617168/slides_Ngadiuba.pdf … [Video] https://cds.cern.ch/record/2315491 https://hls-fpga-machine-learning.github.io/hls4ml/ HLS4ML, BOOST 2018, Jul 19, 2018 PDF https://indico.cern.ch/event/649482/contributions/2993319/attachments/1689369/2717753/HLS4ML_BOOST2018_Ngadiuba.pdf …pic.twitter.com/p5s0Q02xTV
=> "Design Trade-offs for Machine Learning Solutions on Reconfigurable Devices (FPGA)", Michaela Blott, Xilinx Research, Keynote, ASAP 2018, Jul 12, 2018 PDF https://asap18.necst.it/docs/ASAP2018_Blott.pdf … https://scholar.google.com/citations?hl=en&user=Ny8sOl8AAAAJ&view_op=list_works&sortby=pubdate … FINN, Jan 2017 https://twitter.com/ogawa_tter/status/824867533534556161 … Mar, 2017 https://twitter.com/ogawa_tter/status/839461446278758400 …pic.twitter.com/5jENpGRno6
=> "Energy aware transprecision computing FPGA programming using arbitrary precision data-types", Dionysios Diamantopoulos, IBM Res Zurich, NiPS Summer School 2018, Jul 2018 PDF http://www.nipslab.org/sites/nipslab.org/files/OPRECOMP_Perugia_summer_school_July_2018_Diamantopoulos_IBM.pdf … 84 pp ISCAS 2018 https://ieeexplore.ieee.org/document/8351755/ … CAPI SNAP https://twitter.com/ogawa_tter/status/1022104968826781696 …pic.twitter.com/sY1nytW37i
=> "Compression of Deep Neural Networks (for FPGAs & Trigger)", Fermi National Accelerator Lab, Apr 27, 2018 PDF https://indico.cern.ch/event/714134/contributions/2961605/attachments/1641484/2621597/MIT_NetworkCompression_26Apr2018.pdf … "Why compress neural networks?" Many different techniques / implementations HLS4ML, CERN, Apr 2018 & Jul 2018 https://twitter.com/ogawa_tter/status/1024284910117244929 …pic.twitter.com/MPtO5yHGnw
=>
Accelerating SqueezeNet on FPGA, M. Arora & S. Lanka https://lankas.github.io/15-618Project/
Spring 2017 Student Projects, CMU 15-418/618 Parallelism Competition http://15418.courses.cs.cmu.edu/spring2017/projects …
via @marsee101
Compression of Deep Neural Networks (for FPGAs & Trigger), Apr 2018https://twitter.com/ogawa_tter/status/1025220782882873344 …
=> "MIRU MIRU わかる GAN "、MIRU 2018 チュートリアル、2018年8月5日 https://www.slideshare.net/TomohiroTakahashi2/miru-miru-gan … FlexiGAN: An End-to-End Solution for FPGA Acceleration of GANs, FCCM 2018 https://twitter.com/ogawa_tter/status/1018514384518123520 … GANAX: A Unified SIMD-MIMD Acceleration for GANs, ISCA 2018 https://twitter.com/ogawa_tter/status/1018515569312595969 …pic.twitter.com/k3v7CYugnc
=> 「高効率・高速処理を可能とするAIチップ・次世代コンピューティングの技術開発／【研究開発項目〔1〕】革新的AIエッジコンピューティング技術の開発」に係る実施体制の決定、NEDO、平成30年8月10日 http://www.nedo.go.jp/koubo/IT3_100047.html … 研究開発枠：7 先導調査研究枠：3 公募情報 https://twitter.com/ogawa_tter/status/998848148964442113 …pic.twitter.com/lD8i6Isr4Z
=> Sophon, Bitmain https://www.sophon.ai/ "BITMAIN launches SOPHON Tensor Processors and AI Solutions", Nov 11, 2017 https://sophon.ai/post/12.html BM1680 Datasheet, Oct 2017 https://sophon-file.bitmain.com.cn/sophon-prod/drive/BM1680_Datasheet%20V1.0.pdf … SC1 White Paper, Oct 2017 https://sophon-file.bitmain.com.cn/sophon-prod/drive/SC1_White%20Paper%20V1.0.pdf … Roadmap https://twitter.com/david_schor/status/1028285124629213184 …pic.twitter.com/txTcucr12q
=> BITMAIN, is officially joining ONNX, Apr 8, 2018 https://www.sophon.ai/post/15.html BITMAIN partners with Skymizer for ONNX, Jun 4, 2018 https://skymizer.com/news/onnc_june https://skymizer.com/ Compiler company Preview Release: ONNC, Aug 6, 2018 https://www.sophon.ai/post/32.html https://twitter.com/ogawa_tter/status/1028297399880404992 …
=> "How Intel Xeon Changes (AMD EPYC Infinity Fabric Topology, NUMA) Impacted Single Root Deep Learning Servers", Aug 1, 2018 https://www.servethehome.com/how-intel-skylake-sp-changes-impacted-single-root-pcie-deep-learning-servers/ … PCIe topology becomes more important "why the new generation of CPUs creates a problem for the newest architectures." 確かに…
=>
"EPYCマシンの検証(2) - NUMAノードをまたぐメモリアクセス速度"、 @satoru_takeuchi 、サイボウズエンジニアのブログ、2018年3月29日 http://blog.cybozu.io/entry/2018/03/29/170000 …
NUMA構成でのメモリアクセス速度の違いについて実測よる報告。PCIeアクセスでも同様に差がでるはず…https://twitter.com/ogawa_tter/status/1028354078558978048 …
=> "Optical Artificial Neural Network", Yichen Shen, Apr 26, 2018 PPTX https://indico.cern.ch/event/714134/contributions/2976665/attachments/1640802/2620026/yichen_presentation.pptx … Shen Yichen https://www.shenyichen.org/ MIT News, Jun 12, 2017 http://news.mit.edu/2017/new-system-allows-optical-deep-learning-0612 … Nature Photonics, Jun 2017 https://www.nature.com/articles/nphoton.2017.93 … PhD Thesis, 2016 https://dspace.mit.edu/handle/1721.1/104534 …pic.twitter.com/H6S93I9Ilb
=> 「次世代人工知能・ロボット中核技術開発における成果の効果的な訴求方法の調査・検討」実施体制の決定 、NEDO、平成30年8月14日 http://www.nedo.go.jp/koubo/CD3_100126.html … 市場予測 マッチングイベント等の企画・立案・運営 対外PRに必要となる資材作成や情報収集等 ⇒研究開発成果の効果的な訴求方法の調査・検討
=> "Efficient Edge Computing for Deep Neural Networks and Beyond", V. Sze, MIT, MSR Res Faculty Summit, Aug 1, 2018 https://www.microsoft.com/en-us/research/uploads/prod/2018/08/Free_Inference_and_Instant_Training_Breakthroughs_and_Implications_Vivienne_Sze_Matthai_Philipose.pdf … Jun 2018 https://twitter.com/ogawa_tter/status/1011965499893170176 … NetAdapt, arXiv, Apr 9 2018 https://arxiv.org/abs/1804.03230 Eyeriss v2, arXiv, Jul 10 2018 https://arxiv.org/abs/1807.07928 pic.twitter.com/Ibi4ub87vD
=> "Medusa: A Scalable Interconnect for Many-Port DNN Accelerators and Wide DRAM Controller Interfaces", arXiv, Jul 11, 2018 (FPL 2018, Aug 27, 2018) https://arxiv.org/abs/1807.04013 Yongming Shen http://compas.cs.stonybrook.edu/~yoshen/ Peter Milder, Stony Brook University https://www.petermilder.com/ pic.twitter.com/T7XqgDiNFi
=> Gravity: Efficient Training on Very Large Corpora via Gramian Estimation, N Mayoraz, Google, Extreme Classification, Dagstuhl Seminar, Jul, 2018 https://materials.dagstuhl.de/files/18/18291/18291.NicolasMayoraz.Slides.pdf … W. Krichene, N. Mayoraz, et al., Google, arXiv, Jul 18 2018 https://arxiv.org/abs/1807.07187 http://walid.krichene.net/ pic.twitter.com/TVESb8tmBP
=> 「人工知能技術適用によるスマート社会の実現」に係る実施体制の決定、NEDO、平成30年8月16日 http://www.nedo.go.jp/koubo/CD3_100108.html … 委託予定先 PDF http://www.nedo.go.jp/content/100882003.pdf … 「AI技術による超スマート社会」実現プロジェクトが始動、2018年8月16日 http://www.nedo.go.jp/news/press/AA5_101009.html … テーマ PDF http://www.nedo.go.jp/content/100882324.pdf …pic.twitter.com/8VYYbsHhM3
=> "Intel buys deep-learning startup http://Vertex.AI to join its Movidius unit", Apr 16, 2018 https://techcrunch.com/2018/08/16/intel-buys-deep-learning-startup-vertex-ai-to-join-its-movidius-unit/ … http://vertex.ai/ Vertex AI is now part of Intel's AI Products Group. Intel plans to continue developing PlaidML PlaidMLhttps://github.com/plaidml/plaidml
=> "Two Startups Use Processing in Flash Memory for AI at the Edge", AUg 13, 2018 https://spectrum.ieee.org/tech-talk/computing/embedded-systems/two-startups-use-processing-in-flash-memory-for-ai-at-the-edge … Mythic AI https://twitter.com/ogawa_tter/status/996898646485422081 … Syntiant https://www.syntiant.com/ CTO https://scholar.google.com/citations?hl=ja&user=kWTef0wAAAAJ&view_op=list_works&sortby=pubdate … COO, VP SW, VP of HW were at Broadcom May 9, 2018 https://www.syntiant.com/news/ai-semiconductor-company-syntiant-demonstrates-analog-neural-network …
=> "SK Telecom Deploys Xilinx FPGAs for AI Acceleration, Achieves 5X Performance/16X Performance-per-watt over GPUs", Aug 15, 2018 https://www.xilinx.com/news/press/2018/sk-telecom-deploys-xilinx-fpgas-for-ai-acceleration-achieves-5x-performance-16x-performance-per-watt-over-gpus.html … "SK Telecom 社、AI アクセラレータとしてザイリンクスの FPGA を採用、GPU 比で …"、2018年8月16日 https://japan.xilinx.com/news/press/2018/sk-telecom-deploys-xilinx-fpgas-for-ai-acceleration.html …
=> DARPA ERI Materials and Integration ("Page 3") – Framework for Novel Compute (FRANC), DARPA ERI Summit, Jul 24, 2018 PDF https://uploads-ssl.webflow.com/5a749b2fa5fde0000189ffc0/5b6b3d0b9e9b6f2969393502_20180724_1230_FRANC.pdf … "FRANC will enhance Memory-Centric Computing Architecture" Micron/UIUC (DIMA) Ref https://twitter.com/ogawa_tter/status/1024670739016736768 … https://twitter.com/ogawa_tter/status/1031166652199497729 …pic.twitter.com/x8vjCP3CMS
=> Xilinx: Vision for the Future of Computing at Hot Chips 2018, Aug 20, 2018 https://www.xilinx.com/news/media-kits/xilinx-unveils-its-vision-for-the-future-of-computing-details-new-programmable-engine-fabric-and-multiple-ai-technologies.html#releases … PDF Xilinx DNN Processor https://www.xilinx.com/content/dam/xilinx/imgs/press/media-kits/Xilinx_DNN_Processor_Hot_Chips_2018.pdf … HW/SW Programmable Engine https://www.xilinx.com/content/dam/xilinx/imgs/press/media-kits/Xilinx_SWPE_Hot_Chips_2018.pdf … Architectures for Accelerating Deep Neural Networks https://www.xilinx.com/content/dam/xilinx/imgs/press/media-kits/Deep_Learning_and_Computer_Architectures_Hot_Chips_2018.pdf …pic.twitter.com/5n1zvfO37a
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
