[B! TensorFlow] z_dogmaã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

z_dogma id:z_dogma

TensorFlowã«é–¢ã™ã‚‹z_dogmaã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (10)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

TensorFlowã§ç‰©ä½“é ˜åŸŸäºˆæ¸¬ï¼ˆRegion Proposalï¼‰ã‚’è©¦ã—ã¦ã¿ã‚‹ | Workpiles
æœ€è¿‘ã€ã¡ã‚‡ã£ã¨ã€ŽSSD:Single Shot MultiBox Detectorã€ã‚’å‹‰å¼·ä¸ã§ã™ã€‚ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ä½¿ã£ãŸä¸€èˆ¬ç‰©ä½“æ¤œå‡ºã®ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã§ã€å†™çœŸã®â€œã©ã“ã«â€â€œãªã«ãŒâ€å†™ã£ã¦ã„ã‚‹ã‹ã‚’æŽ¨æ¸¬ã™ã‚‹ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã§ã™ã€‚ ä»Šå›žã¯ã€TensorFlowã§ç‰©ä½“é ˜åŸŸäºˆæ¸¬ã‚’è©¦ã—ã¦ã¿ã¾ã—ãŸã€‚SSDã«ã¤ã„ã¦ã‚‚æ›¸ã„ã¦ã„ã¾ã™ãŒã€ä»Šå›žè©¦ã—ãŸã“ã¨ã¨ã¯ã»ã¨ã‚“ã©é–¢ä¿‚ãªã‹ã£ãŸã‚Šã—ã¾ã™ï¼ˆå‚è€ƒã«ã¯ã—ã¦ã„ã¾ã™ãŒâ€¦SSDé›£ã—ã„ï¼‰ ä¸€èˆ¬ç‰©ä½“æ¤œå‡ºã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã«ã¤ã„ã¦ ä¸€èˆ¬ç‰©ä½“æ¤œå‡ºã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã«ã¤ã„ã¦ã¯ã€ä¸‹è¨˜ãƒ–ãƒã‚°ã§åˆ†ã‚Šæ˜“ãç´¹ä»‹ã•ã‚Œã¦ã„ã¾ã™ã€‚ Deep Learningã«ã‚ˆã‚‹ä¸€èˆ¬ç‰©ä½“æ¤œå‡ºã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã®ç´¹ä»‹ ã©ã®ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã‚‚arXivã«è«–æ–‡ãŒä¸ŠãŒã£ã¦ã„ã¾ã™ã€‚ SSDã‹ã‚‰èªã¿å§‹ã‚ã‚‹ã¨ã€Faster R-CNNã€Fast R-CNNã€MultiBoxã€YOLOè¾ºã‚Šã¯èŠ‹ã¥ã‚‹ã§èªã‚€å¿…è¦ãŒå‡ºã¦ãã‚‹ã®ã§ã€å…¨ã¦ã‚’ç†è§£ã™ã‚‹ã®ã¯çµæ§‹å¤§å¤‰ã ã¨æ€ã†ã‚
z_dogma 2017/11/12
SSD

ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ

TensorFlow
ãƒªãƒ³ã‚¯
TensorFlowã«define by runï¼ˆTensorFlow Eagerï¼‰ãŒã‚„ã£ã¦ããŸ - ã®ã‚“ã³ã‚Šã—ã¦ã„ã‚‹ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã®æ—¥è¨˜
çš†ã•ã‚“ã“ã‚“ã«ã¡ã¯ ãŠå…ƒæ°—ã§ã™ã‹ã€‚ä»Šå›žã®ä¸‰é€£ä¼‘ã¯äºŒéƒŽé£Ÿã¹ã¦æº€è¶³ã—ã¾ã—ãŸã€‚ Chainerã«ã¯ã˜ã¾ã‚Šã€PyTorchãªã©define by runã§ ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚’è¨ˆç®—ã™ã‚‹ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ãŒã‚ã‚Šã¾ã™ã€‚ ã“ã®define by runã¯éžå¸¸ã«RNNç³»ã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚’æ›¸ãéš›ã«é‡å®ã—ã¦ã„ã¾ã™ã€‚ ãã—ã¦ã€10æœˆæœ«ã«TensorFlowã‹ã‚‰ã‚‚define by runç”¨ã® ã‚¤ãƒ³ã‚¿ãƒ¼ãƒ•ã‚§ãƒ¼ã‚¹ãŒè©¦é¨“çš„ã«æä¾›ã•ã‚Œã¾ã—ãŸã€‚ãã‚ŒãŒTensorFlow Eagerã§ã™ã€‚ â€»define by runãŒä¸æ˜Žãªæ–¹ã¯ã“ã¡ã‚‰ã¸ s0sem0y.hatena blog.com TensorFlow Eager TensorFlow Eagerã¯æ¬¡ã®å…¬å¼ã®è¨˜äº‹ã§ç´¹ä»‹ã•ã‚Œã¦ã„ã¾ã™ã€‚ ãŸã ã—ã€ã“ã®æ©Ÿèƒ½ã¯Previewæ®µéšŽã§ã™ã€‚ é€šå¸¸ä½¿ã†å ´é¢ã§ã¯å•é¡Œãªã„ã¨æ€ã„ã¾ã™ãŒã€å¿µã®ãŸã‚ä½•ã‹èµ·ã“ã£ã¦ã‚‚å•é¡Œãªã„ç’°å¢ƒã§åˆ©ç”¨ã—ã¦ãã ã•
z_dogma 2017/11/06
TensorFlow

define by run
ãƒªãƒ³ã‚¯
éŽæ¸¡æœŸã«ã‚ã‚‹TensorFlowã¨Kerasã®é–¢ä¿‚ã‚’èª¿ã¹ã‚‹ - Qiita
ï¼ˆ å¤‰æ•°åã‚³ãƒ³ãƒˆãƒãƒ¼ãƒ«ã®é•ã„ ã«ã¤ã„ã¦è¿½è¨˜ã—ã¾ã—ãŸï¼Žï¼‰ ã¯ã˜ã‚ã« TensorFlow Dev Summitãªã©ã§æƒ…å ±ãŒãƒªãƒªãƒ¼ã‚¹ã•ã‚Œã¦ã„ã¾ã™ãŒï¼ŒTensorFlowã¨Kerasã®çµ±åˆãŒé€²ã‚ã‚‰ã‚Œã¦ã„ã¾ã™ï¼ŽKeras Blog - Introducing Keras 2 ã‹ã‚‰å¼•ç”¨ã—ã¾ã™ï¼Ž Keras is best understood as an API specification, not as a specific codebase. In fact, going fowards there will be two separate implementations of the Keras spec: the internal TensorFlow one, available as tf.keras, written in pure TensorFlow and deeply comp
z_dogma 2017/11/06
TensorFlow
ãƒªãƒ³ã‚¯
ã€å…¥é–€è€…å‘ã‘è§£èª¬ã€‘ç•³ã¿è¾¼ã¿å‡¦ç†å…¥é–€(TensorFlowã§èª¬æ˜Ž) - Qiita
å…¥é–€è€…å‘ã‘ã«ã€Œç•³ã¿è¾¼ã¿ã€å‡¦ç†ã«ã¤ã„ã¦è§£èª¬ TensorFlowã‚¨ã‚ã‚¹ãƒ‘ãƒ¼ãƒˆå‘ã‘ãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«Deep MNIST for Expertsã‚’å®Ÿè¡Œã—ã¦ã€ã¾ãšã‚ã‹ã‚Šã«ãã„ã®ãŒã€Œç•³ã¿è¾¼ã¿ã€å‡¦ç†ã§ã—ãŸã€‚æ–‡ç³»å’ã‹ã‚‰ã™ã‚‹ã¨ä¸€èˆ¬ç”¨èªžã§ã‚‚ãªã„ã§ã™ã—ã€å°‘ã—æ™‚é–“ãŒã‹ã‹ã‚Šã¾ã—ãŸãŒã€åˆ†ã‹ã‚Œã°ç°¡å˜ã§ã™ã€‚åŸºæœ¬çš„ã«ã¯å››å‰‡æ¼”ç®—ãŒã§ãã‚Œã°ç†è§£ã§ãã¾ã™ã€‚ ã€å…¥é–€è€…å‘ã‘è§£èª¬ã€‘TensorFlowãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«Deep MNISTè¨˜äº‹ã§ã€ç•³ã¿è¾¼ã¿ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®å‡¦ç†ã‚’è§£èª¬ã—ã¦ã„ã¾ã™ã®ã§å‚è€ƒã«ã—ã¦ãã ã•ã„ã€‚ ã¾ãŸã€ã€Œã€å…¥é–€è€…å‘ã‘è§£èª¬ã€‘ãƒ—ãƒ¼ãƒªãƒ³ã‚°å‡¦ç†å…¥é–€(TensorFlowã§èª¬æ˜Ž)ã€ã®è¨˜äº‹ã§ãƒ—ãƒ¼ãƒªãƒ³ã‚°ã«ã¤ã„ã¦ã‚‚è§£èª¬ã—ã¦ã„ã¾ã™ã€‚ â€»TensorBoardã§å‡ºåŠ›ã—ãŸç”»åƒã‚’å‚è€ƒã«è¼‰ã›ã¾ã—ãŸ(2017/7/27) æ¦‚èª¬ æ¦‚è¦ ã€Œã€å…¥é–€è€…å‘ã‘è§£èª¬ã€‘TensorFlowãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«Deep MNISTã€ã®è¨˜äº‹ã§è§£èª¬ã—ã¾ã—ãŸãŒã€ã€Œç•³ã¿è¾¼
z_dogma 2017/10/22
TensorFlow
ãƒªãƒ³ã‚¯
TensorFlowã®ãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«ã‚’é€šã—ã¦ã€äººå·¥çŸ¥èƒ½ã®åŽŸç†ã«ã¤ã„ã¦å¦ç¿’ã™ã‚‹ - Qiita
Deleted articles cannot be recovered. Draft of this article would be also deleted. Are you sure you want to delete this article? ä»¥ä¸‹ã®ã“ã¨ã«ã¤ã„ã¦çŸ¥ã‚ŠãŸã„äººå‘ã‘ã®è¨˜äº‹ã€‚ â‘ æ©Ÿæ¢°å¦ç¿’ã®åˆ©ç‚¹ã¨æ¬ ç‚¹ã€TensorFlowã®åˆ©ç‚¹ã¨æ¬ ç‚¹ â‘¡ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®åŽŸç†ã«ã¤ã„ã¦ã®ç°¡æ½”ãªèª¬æ˜Ž (ã‚ªãƒ³ãƒ©ã‚¤ãƒ³æ›¸ç±ï¼šhttp://nnadl-ja.github.io/nnadl_site_ja/index.html ã‚’ã¾ã¨ã‚ãŸã‚‚ã®) â‘¢TensorFlowã®ãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«ã«æ²¿ã£ã¦ãƒ—ãƒã‚°ãƒ©ãƒ ä½œæˆã€ãã®å®Ÿè¡Œçµæžœ ä»¥ä¸‹ã€å‚è€ƒã«ã—ãŸã‚µã‚¤ãƒˆ 1æ¬¡é–¢æ•°ã®è¿‘ä¼¼ï¼š https://www.tensorflow.org/versions/master/get_started/index.html h
z_dogma 2017/10/22
TensorFlow

ã‚ã¨ã§èªã‚€
ãƒªãƒ³ã‚¯
è‡ªç„¶è¨€èªžç³»ã®ãƒ‡ãƒ¼ã‚¿ã§LSTMç†è§£ã—ã‚ˆã†ã¨ã™ã‚‹ã¨é›£ã—ã„ã®ã§ã€MNISTãƒ‡ãƒ¼ã‚¿ã§LSTMã®åŸºç¤Žã‚’è§£èª¬ - Qiita
å¸«èµ°ã«èª¿åä¹—ã£ã¦Tensorflow Advent calendar 2016ç™»éŒ²ã—ã¦å®Œå…¨ã«ãƒ’ãƒ¼ãƒ’ãƒ¼è¨€ã„ãªãŒã‚‰ä»Šå›žã®å†…å®¹ã‚’æ›¸ãã¾ã—ãŸã€‚ ãŠãã‚‰ãå¾Œã‹ã‚‰èªã¿ã‚„ã™ã„ã‚ˆã†ã«è¿½è¨˜ã‚„æ›´æ–°ã‚‚ã™ã‚‹ã¨æ€ã„ã¾ã™ã€‚8æ—¥ç›®ã§ã™ã€‚ã©ã†ã‚‚ã€‚ ã“ã®è¨˜äº‹ã¯Tensorflowã®ãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«ã‚’èªã‚“ã§ãªã‚“ã¨ãªãã®æ©Ÿæ¢°å¦ç¿’çš„ãªç†è§£ãŒã‚ã‚‹æ–¹é”ãŒãŠãã‚‰ãå¯¾è±¡ã§ã™ã€‚ ãã‚Œã‚‰ã«é–¢é€£ã—ãŸè§£èª¬ã‚‚ä¸€å¿œæ›¸ã„ã¦ã¾ã™ã€‚ ãƒ»ãƒ“ã‚®ãƒŠãƒ¼ã®è§£èª¬ : ç‰¹ã«ãƒ—ãƒã‚°ãƒ©ãƒžãƒ¼ã§ã‚‚ãƒ‡ãƒ¼ã‚¿ã‚µã‚¤ã‚¨ãƒ³ãƒ†ã‚£ã‚¹ãƒˆã§ã‚‚ãªã„ã‘ã©ã€Tensorflowã‚’1ãƒ¶æœˆè§¦ã£ãŸã®ã§è¶…åˆ†ã‹ã‚Šã‚„ã™ãè§£èª¬ ãƒ»ã‚¨ã‚ã‚¹ãƒ‘ãƒ¼ãƒˆã®è§£èª¬ : Tensorflowã‚’ï¼’ãƒ¶æœˆè§¦ã£ãŸã®ã§"æ‰‹æ›¸ãã²ã‚‰ãŒãª"ã®è˜åˆ¥95.04%ã§ç•³ã¿è¾¼ã¿ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚’ã‚ã‹ã‚Šã‚„ã™ãè§£èª¬ æ©Ÿæ¢°å¦ç¿’ãŒã©ã‚“ã©ã‚“ç››ã‚Šä¸ŠãŒã£ã¦ã¾ã™ãã€‚ã€€ã¿ãªã•ã‚“æ¥½ã—ãã†ã€‚ ãã‚“ãªã‚ã‘ã§ä»Šæ›´word2vecãªã‚“ã¦ã‚„ã£ã¦ã‚‚...æ„Ÿã‚‚ã‚ã‚‹ã®ã§ã€ã™ã£é£›ã°ã—ã¦Re
z_dogma 2017/10/10
æ©Ÿæ¢°å¦ç¿’

TensorFlow

ã‚ã¨ã§èªã‚€
ãƒªãƒ³ã‚¯
TensorFlowã®ãƒ‡ãƒ¼ã‚¿ãƒ•ã‚©ãƒ¼ãƒžãƒƒãƒˆTFRecordã®æ›¸ãè¾¼ã¿ã¨èªã¿è¾¼ã¿æ–¹æ³•
TFRecordã‚’ä½¿ç”¨ã™ã‚‹ç†ç”± TFRecordã®ä½œã‚Šæ–¹ Exampleãƒ¬ã‚³ãƒ¼ãƒ‰ã¨SequenceExampleãƒ¬ã‚³ãƒ¼ãƒ‰ Fashion MNISTã‚’TFRecordåŒ–ã—ã¦ã¿ã‚‹ TFRecordã®ä¸èº«ã‚’ç¢ºèªã™ã‚‹æ–¹æ³• TFRecordã®èªã¿è¾¼ã¿æ–¹ å®Ÿéš›ã«å®Ÿè£…ã—ã¦ã¿ã‚‹ TFRecordã‚’ä½¿ç”¨ã™ã‚‹ ã¾ã¨ã‚ å‚è€ƒ TensorFlowãŒæŽ¨å¥¨ã—ã¦ã„ã‚‹TFRecordã¨ã„ã†ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆã®ãƒ•ã‚©ãƒ¼ãƒžãƒƒãƒˆãŒã‚ã‚Šã¾ã™ã€‚ TFRecordã‚’ä½¿ã„ã“ãªã›ã‚‹ã‚ˆã†ã«ãªã‚‹ã¨ã€å¤§è¦æ¨¡ãªãƒ‡ãƒ¼ã‚¿ã‚’åŠ¹çŽ‡çš„ã«å¦ç¿’ã§ãã‚‹ã‚ˆã†ã«ãªã‚‹ã“ã¨ãŒã‚ã‚Šã¾ã™ã€‚ æœ¬è¨˜äº‹ã§ã¯ã€TFRecordã®ä½¿ã„æ–¹ã‚’ãƒžã‚¹ã‚¿ãƒ¼ã§ãã‚‹ã‚ˆã†ã«èªã¿æ›¸ãã™ã‚‹æ–¹æ³•ã‚’è§£èª¬ã—ã€å®Ÿéš›ã«QueueRunnerã‚’ä½¿ã£ãŸå®Ÿè£…ã‚’ã—ã¦ã¿ã¾ã™ã€‚ TFRecordã‚’ä½¿ç”¨ã™ã‚‹ç†ç”± TFRecordã®ä¸èº«ã¯Protocol Bufferã¨ã„ã†ãƒã‚¤ãƒŠãƒªãƒ•ã‚©ãƒ¼ãƒžãƒƒãƒˆã§ã™ã€‚ä¸€åº¦TFRecordã‚’ä½œæˆã™ã‚‹ã“
z_dogma 2017/10/08
TensorFlow
ãƒªãƒ³ã‚¯
æ–°ã—ã„ TensorBoard API ã§æ©Ÿæ¢°å¦ç¿’ã®ç‹¬è‡ªã®å¯è¦–åŒ–ã‚’ãƒ“ãƒ«ãƒ‰
.app 1 .dev 1 #11WeeksOfAndroid 13 #11WeeksOfAndroid Android TV 1 #Android11 3 #DevFest16 1 #DevFest17 1 #DevFest18 1 #DevFest19 1 #DevFest20 1 #DevFest21 1 #DevFest22 1 #DevFest23 1 #hack4jp 3 11 weeks of Android 2 A MESSAGE FROM OUR CEO 1 A/B Testing 1 A4A 4 Accelerator 6 Accessibility 1 accuracy 1 Actions on Google 16 Activation Atlas 1 address validation API 1 Addy Osmani 1 ADK 2 AdMob 32 Ads
z_dogma 2017/10/05
TensorFlow

API
ãƒªãƒ³ã‚¯
ã€Day-17ã€‘DeepLearningç³»ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã€ã€ŽKerasã€ã®ä½¿ã„æ–¹ã¾ã¨ã‚(2.xå¯¾å¿œç‰ˆ) - ãƒ—ãƒã‚¯ãƒ©ã‚·ã‚¹ãƒˆ
ã€æœ€çµ‚æ›´æ–° : 2017.12.17ã€‘ â€»ä»¥å‰æ›¸ã„ãŸè¨˜äº‹ãŒObsoleteã«ãªã£ãŸãŸã‚ã€2.xã§ãã¡ã‚“ã¨å‹•ãã‚ˆã†ã«æ›¸ãç›´ã—ã¾ã—ãŸã€‚ ãƒ‡ãƒ¼ã‚¿åˆ†æžã‚¬ãƒå‹‰å¼·ã‚¢ãƒ‰ãƒ™ãƒ³ãƒˆã‚«ãƒ¬ãƒ³ãƒ€ãƒ¼ 17æ—¥ç›®ã€‚ 16æ—¥ç›®ã«ã€1ã‹ã‚‰ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã‚’æ›¸ãã¾ã—ãŸã€‚ ãã‚Œã¯ãã‚Œã§ãƒ‡ãƒ¼ã‚¿ã®æµã‚Œã ã¨ã‹ã€æ´»æ€§åŒ–é–¢æ•°ã®åƒãã ã¨ã‹å¾—ã‚‹ã‚‚ã®ã¯å¤šã‹ã£ãŸã®ã§ã™ãŒã€Kerasã¨è¨€ã†ã‚‚ã®ã‚’ä½¿ã£ã¦ã¿ã¦ã€ä½•ã¦ç´ æ™´ã‚‰ã—ã„ã‚“ã ã¨æ„Ÿå‹•ã—ã¦ã—ã¾ã„ã¾ã—ãŸ ä»Šã¾ã§è‹¦åŠ´ã—ã¦æ•°åè¡Œæ›¸ã„ã¦ã„ãŸã‚‚ã®ãŒã€ã‚ãšã‹3è¡Œã§æ›¸ã‘ã‚‹ï¼ æ£ç›´ã€ã‚¹ã‚¯ãƒ©ãƒƒãƒã§æ›¸ãæ„å‘³ã£ã¦ã€ç†è§£ã«ã¯ã„ã„ã‘ã©ç ”ç©¶ã‚„åˆ†æžã«ã¯å¿…è¦ã‚ã‚“ã¾ãªã„ã‚“ã§ã™ã‚ˆãã€‚è»Šè¼ªã®å†ç™ºæ˜Žã«ãªã‚‹ã—ã€‚ ã¨è¨€ã†ã‚ã‘ã§ã€ä½¿ãˆã‚‹ã‚‚ã®ã¯ã©ã‚“ã©ã‚“ä½¿ã£ã¦ã„ã“ã†ã‚¹ã‚¿ãƒ³ã‚¹ã§ã€ä»Šæ—¥ã¯Kerasã®ç´¹ä»‹ã§ã™ï¼ Tutorial+æ°—ã«ãªã£ãŸå¼•æ•°ã‚’æŽ˜ã‚Šä¸‹ã’ã¦è£œè¶³ã®ã‚ˆã†ãªæ„Ÿã˜ã§æ›¸ã„ã¦ã„ã¾ã™ã€‚ ã¡ãªã¿ã«ã€å„éƒ¨ã®ã‚³ãƒ¼ãƒ‰ä»¥ä¸‹ã‚’ã¤ãªãŽåˆã‚ã›ã‚‹ã¨mnistã®åˆ†é¡žå™¨ãŒå‹•ãã‚ˆ
z_dogma 2017/01/08
TensorFlow

DeepLearning
ãƒªãƒ³ã‚¯
TensorFlow Playgroundã¯ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã‚’ç†è§£ã™ã‚‹ã®ã«ãŠã‚¹ã‚¹ãƒ¡ - è¥¿å°¾æ³°å’Œã®ã¯ã¦ãªãƒ€ã‚¤ã‚¢ãƒªãƒ¼
ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®é‡ã¿ã‚„å„ãƒ‹ãƒ¥ãƒ¼ãƒãƒ³ãŒã©ã†ã„ã†å…¥åŠ›ã®æ™‚ã«ç™ºç«ã™ã‚‹ã®ã‹ãŒã€å¦ç¿’ã—ã¦ã„ãéŽç¨‹ã§å„æ™‚åˆ»å¯è¦–åŒ–ã•ã‚Œã¦ã¨ã¦ã‚‚è‰¯ã„æ•™æã§ã™ã€‚ http://playground.tensorflow.org/ ã†ãšã¾ãã®ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆã«é–¢ã—ã¦ã€Œä¸é–“å±¤ãŒ1å±¤ã—ã‹ãªã„ã¨ã†ãšã¾ãï¼ˆç·šå½¢éžåˆ†é›¢ãªå•é¡Œï¼‰ã¯è§£ã‘ãªã„ã€ã¨ã„ã†èª¤è§£ãŒã‚ã‚‹ã‚ˆã†ãªã®ã§ã€ã¾ãšã¯1å±¤ã§ã§ãã‚‹ã¨ã„ã†çµµã‚’ç´¹ä»‹ã€‚ãªãŠåƒ•ã®ã‚¿ã‚¤ãƒ ãƒ©ã‚¤ãƒ³ä¸Šã§ã¯ id:a2c ãŒåƒ•ã‚ˆã‚Šå…ˆã«æ°—ä»˜ã„ã¦ã„ãŸã“ã¨ã‚’åèª‰ã®ãŸã‚ã«è¨€åŠã—ã¦ãŠãã¾ã™ã€‚ ã§ã€ã˜ã‚ƒã‚ã‚ˆãè¨€ã‚ã‚Œã‚‹ã€Œç·šå½¢éžåˆ†é›¢ãªå•é¡ŒãŒè§£ã‘ãªã„ã€ã£ã¦ã®ã¯ã©ã†ã„ã†ã“ã¨ã‹ã€‚ãã‚Œã¯ã“ã‚“ãªå•é¡Œè¨å®šã€‚å…¥åŠ›ã«é©å½“ãªä¿‚æ•°ã‚’æŽ›ã‘ã¦è¶³ã—åˆã‚ã›ãŸã ã‘ã§ã¯é©åˆ‡ãªå¢ƒç•Œã‚’ä½œã‚‹ã“ã¨ãŒã§ãã¾ã›ã‚“ã€‚ ã“ã†ã„ã†ã‚±ãƒ¼ã‚¹ã§ã¯ä¸é–“å±¤ã‚’è¿½åŠ ã™ã‚‹ã¨ã€ä¸é–“å±¤ãŒå…¥åŠ›ã®éžç·šå½¢ãªçµ„ã¿åˆã‚ã›ã‚’æ‹…å½“ã—ã¦ãã‚Œã‚‹ãŠã‹ã’ã§è§£ã‘ãªã‹ã£ãŸå•é¡ŒãŒè§£ã‘ã‚‹ã‚ˆã†ã«ãªã‚Šã¾ã™ã€‚ 1ã¤ç›®ã®ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆã§ã¯ç‰¹å¾´é‡ã®
z_dogma 2016/04/15
TensorFlow

æ©Ÿæ¢°å¦ç¿’

deepLearning
ãƒªãƒ³ã‚¯
1