Neural Networks and TensorFlow - Deep Learning Series [Part 14] by cristi

View this thread on: hive.blog | peakd.com | ecency.com

deep-learning · @cristi · May 16 '18

$22.24

Neural Networks and TensorFlow - Deep Learning Series [Part 14]

Even though we've previously discussed about activations, in this lesson we're going to look into it specifically for convolutional neural networks in TensorFlow. 

So, in brief, in this short lesson:

- we discuss the types of activations commonly used for CNNs in TensorFlow
- pooling layers and how to implement them
- dropout, what it is used for, and how to implement it. 

If you've been watching this series, thank you! And I hope you learn something from it, at least as much as I want to convey knowledge about this stuff. 

<center>https://www.youtube.com/watch?v=lN2fFRgA8gw</center>
___
### <center>To stay in touch with me, follow @cristi</center>   
___
[Cristi Vlad](http://cristivlad.com) Self-Experimenter and Author

👍 busy.pay, jaki01, hendrikdegrote, cristi, gavvet, ace108, steemstem, justtryme90, lafona-miner, equ1l1br1um, pharesim, decebal2dac, mrtv2, coinbitgold, jasonstaggers, team-leibniz, dragonslayer109, jens84, curie, meerkat, geke, anwenbaumeister, thecryptofiend, lemouth, chrone, scaredycatguide, alexvan, mandela, fredrikaa, veleje, timsaid, anarchyhasnogods, jjb777, kerriknox, guyfawkes4-20, ngos, sportspodium, alexandraioana26, mitchelljaworski, liberosist, vact, sme, wonderwomancodes, massivevibration, mountain.phil28, planetenamek, africaunited, kryzsec, thevenusproject, brobear1995, de-stem, terrylovejoy, rbm, gra, benleemusic, thinknzombie, dber, dna-replication, jacor, kenadis, nitesh9, croctopus, michelios, nedspeaks, and 57 others

`author`	cristi
`permlink`	neural-networks-and-tensorflow-deep-learning-series-part-14
`category`	deep-learning
`json_metadata`	{"community":"busy","app":"busy/2.4.0","format":"markdown","users":["cristi"],"links":["/@cristi","http://cristivlad.com"],"tags":["deep-learning","science","busy","programming","python"]}
`created`	2018-05-16 12:13:48
`last_update`	2018-05-16 12:13:48
`depth`	0
`children`	3
`last_payout`	2018-05-23 12:13:48
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	18.909 HBD
`curator_payout_value`	3.334 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	733
`author_reputation`	128,305,218,872,904
`root_title`	"Neural Networks and TensorFlow - Deep Learning Series [Part 14]"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	55,999,397
`net_rshares`	4,531,692,357,824
`author_curate_reward`	""

properties (23)vote details (121)

voter	rshares	pct
pharesim	82,324,678,616	0.02%
lafona-miner	105,243,900,737	5%
gavvet	293,419,069,160	2%
dragonslayer109	45,333,763,021	2%
thecryptofiend	21,620,188,415	25%
justtryme90	119,579,703,361	5%
coinbitgold	57,663,704,811	100%
anwenbaumeister	24,694,515,897	0.89%
field	719,969,071	100%
jacor	1,258,174,680	2%
jens84	42,330,901,435	25%
liberosist	5,480,687,175	0.97%
team-leibniz	47,945,366,106	40%
ace108	191,473,691,249	16%
jasonstaggers	54,954,266,393	25%
timsaid	10,916,688,089	2%
cristi	448,372,554,341	100%
scaredycatguide	16,505,557,777	21%
lemouth	20,788,856,547	5%
rjbauer85	278,282,780	5%
anarchyhasnogods	10,332,951,687	3%
geke	38,874,991,012	50%
meerkat	40,354,902,147	0.97%
mrtv2	57,892,219,477	100%
curie	41,124,583,766	0.89%
busy.pay	1,180,850,336,201	7.63%
mitchelljaworski	5,725,975,758	25%
hendrikdegrote	522,372,459,310	0.89%
jaki01	555,113,453,361	25%
vact	4,891,198,634	0.89%
steemstem	119,864,904,263	5%
decebal2dac	79,296,305,960	100%
j3dy	105,008,038	0.02%
brobear1995	1,969,295,446	100%
foundation	622,105,660	5%
alexvan	13,624,335,569	20%
the-devil	590,405,779	5%
thevenusproject	2,518,843,421	5%
dna-replication	1,263,108,327	5%
mandela	11,902,492,152	2%
borislavzlatanov	386,575,523	5%
michelios	1,060,707,392	0.13%
moksamol	211,001,949	0.44%
thatsweeneyguy	65,611,498	0.44%
kryzsec	2,747,791,830	5%
nedspeaks	963,889,674	5%
fredrikaa	11,364,194,820	2.5%
robertvogt	138,100,195	0.44%
locikll	424,487,046	1.78%
dber	1,290,858,759	5%
aboutyourbiz	211,446,309	0.89%
kerriknox	9,365,349,006	5%
howtostartablog	677,956,201	0.08%
ghostgtr	108,532,250	0.01%
jayna	168,499,856	0.13%
guyfawkes4-20	8,742,740,747	100%
veleje	11,165,351,505	100%
thinknzombie	1,346,438,100	0.44%
nitesh9	1,127,337,985	5%
fancybrothers	635,641,144	1.5%
sportspodium	7,392,042,413	2%
chrone	18,026,914,095	100%
himal	346,906,445	5%
viorel	484,388,722	75%
planetenamek	3,153,713,093	10%
mountain.phil28	3,311,053,734	25%
africaunited	2,868,806,623	2%
ngos	8,036,395,718	2%
modemser	583,030,329	1%
jjb777	9,446,162,927	100%
gabox	182,510,484	0.04%
indy8phish	72,089,821	0.44%
massivevibration	3,537,715,784	5%
onartbali	431,846,179	5%
sme	4,647,617,755	2%
torico	309,680,834	0.07%
simplifylife	757,260,413	2.5%
deeplearning	553,114,804	100%
johnquake	222,963,246	0.44%
mayowadavid	296,724,094	2.5%
happychild	216,387,227	0.44%
imaginedragon	261,724,623	1%
carloserp-2000	902,971,061	5%
rachelsmantra	193,103,174	5%
gra	1,554,627,933	5%
shaff.aff	52,107,827	10%
skycae	252,816,187	0.89%
sireh	354,343,809	0.08%
equ1l1br1um	104,660,334,851	100%
physics.benjamin	166,565,476	5%
kenadis	1,222,897,328	5%
alexandraioana26	6,190,252,180	22%
amavi	780,859,306	1%
robotics101	173,392,211	4%
cgbartow	182,273,323	2.5%
aamin	69,130,216	2.5%
sco	422,276,003	1%
adetola	259,187,656	5%
anikekirsten	255,781,854	0.89%
dysfunctional	252,171,449	2.5%
hadji	218,934,804	5%
terrylovejoy	1,702,631,941	5%
rbm	1,691,896,313	50%
steepup	53,181,009	2%
mikesthoughts	77,281,350	0.44%
hashcash	558,581,647	25%
benleemusic	1,430,935,309	0.08%
croctopus	1,126,673,909	100%
spederson	93,808,402	4.5%
brightideas	927,768,104	2%
de-stem	1,738,651,939	4.5%
jacor-witness	338,873,416	2%
quochuy	97,938,276	0.44%
chloroform	527,308,670	5%
vigna	173,854,326	0.04%
niouton	129,352,937	0.17%
mahfuzzaman	51,891,056	9%
justfell	135,082,690	100%
wonderwomancodes	4,352,065,217	100%
steem.curator	702,973,503	1%
hansmast	81,624,381	0.44%

@adeyemidrey · May 16 '18

This piece as always being interesting ever since e I joined the race of following you,  I have always find it valuable, I really enjoy reading this convolutional neutral network, it widing my knowledge on science related issue, the research work of your is awesome, keep the good work flowing man.

properties (22)

`author`	adeyemidrey
`permlink`	re-cristi-neural-networks-and-tensorflow-deep-learning-series-part-14-20180516t122029146z
`category`	deep-learning
`json_metadata`	{"tags":["deep-learning"],"app":"steemit/0.1"}
`created`	2018-05-16 12:20:33
`last_update`	2018-05-16 12:20:33
`depth`	1
`children`	0
`last_payout`	2018-05-23 12:20:33
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	298
`author_reputation`	29,192,377,423,488
`root_title`	"Neural Networks and TensorFlow - Deep Learning Series [Part 14]"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	56,000,369
`net_rshares`	0

@hashcash · May 16 '18

I think I missed the last part (part13),  but I remember padding. If I remember correctly the output volume keeps on getting smaller for each application of convolution layer. Larger the size of the stride more volume would be lost at the edges.
I think the idea of padding was to start with more volume so we end up retaining most of the original input information.

properties (22)

`author`	hashcash
`permlink`	re-cristi-neural-networks-and-tensorflow-deep-learning-series-part-14-20180516t143638340z
`category`	deep-learning
`json_metadata`	{"tags":["deep-learning"],"app":"steemit/0.1"}
`created`	2018-05-16 14:36:42
`last_update`	2018-05-16 14:36:42
`depth`	1
`children`	0
`last_payout`	2018-05-23 14:36:42
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	366
`author_reputation`	17,433,241,427,492
`root_title`	"Neural Networks and TensorFlow - Deep Learning Series [Part 14]"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	56,020,783
`net_rshares`	0

@olumideolowoyeye · May 16 '18

I Learn quite a lot from your post, no knowledge is lost, no information is useless. What you offer on your blog are what some pay for to learn as private courses and in seminars. If I have any questions I will definitely ask you

properties (22)

`author`	olumideolowoyeye
`permlink`	re-cristi-neural-networks-and-tensorflow-deep-learning-series-part-14-20180516t123716994z
`category`	deep-learning
`json_metadata`	{"tags":["deep-learning"],"app":"steemit/0.1"}
`created`	2018-05-16 12:37:21
`last_update`	2018-05-16 12:37:21
`depth`	1
`children`	0
`last_payout`	2018-05-23 12:37:21
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	229
`author_reputation`	30,997,092,733,643
`root_title`	"Neural Networks and TensorFlow - Deep Learning Series [Part 14]"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	56,002,672
`net_rshares`	0