A Study to Challenge Systematization of Odor - Like Color and Sound by mytechcrunch

science · @mytechcrunch · Mar 9 '17

$4.46

A Study to Challenge Systematization of Odor - Like Color and Sound

<html>
<p><img src="https://3.bp.blogspot.com/-m_vMNhbt_LA/WMAsLpKh2vI/AAAAAAAAGZo/rVEqOvLWjIs6aaEsvO3PzPSHzptFbomUACLcB/s1600/oder-smell.jpg" width="785" height="442"/></p>
<p>For example, "<strong>green</strong>" generally refers to a specific part of the wavelength of light. Similarly, the sound has a frequency. <em>But what about odors?</em> Create a smell chemicals, the interaction is too complicated, simple like a light and sound to smell, but until now there is no way to measure smell of the scent .</p>
<p>But there are known things. Even garlic smell in pee, become the structure of the molecule in the human body is in the receptor original smell.</p>
<p>Having That said, even if you know the combination of the molecule, it is difficult to extract what kind of smell or of shed it is exactly. So <strong>Rockefeller University</strong> and <strong>IBM Thomas J. Watson Research Center</strong>, and by the cooperation of 22 research institutions of the world, has been made information of a molecule model to predict smell based on high-precision. The stakeholders said that this research wants to revolutionize understanding of odor.</p>
<blockquote>"In order to create a new smell, it is necessary to test thousands of odors by a long-time trained specialist, or hundreds of people," IBM research team leader Pablo Meyer says. "But I think that situation will change."</blockquote>
<p>Mr. Meyer organized a contest "<em>DREAM Olfaction Prediction Challenge</em>" to analyze the massive amounts of data so created and attracted 22 teams from research organizations from around the world. 22 teams each creates a smell molecule a prediction to machine learning algorithm, and further, that they are integrated into one, the most accurate smell predictive model ever been build.</p>
<p>In this model, if you input information on the structure and chemical composition of the original molecule of odor, the prediction result of what odor is output. In the research team, 69 of the collected data were used without being used for model construction, and the answer with the prediction model was made with that data. As a result, it was realized <em>high accuracy of 83%</em> .</p>
<p><img src="https://4.bp.blogspot.com/-cAerWtDgxq0/WMAsEH8Te1I/AAAAAAAAGZk/jBa_wE8aSA4lNa7JkgOFosICTxDaotx7QCLcB/s1600/oder-smell-sample.jpg" width="635" height="422"/></p>
<p>This contest model made from, various properties of the substance and various correlations of odor were revealed. For example, a substance containing sulfur was producing "<em>smell like</em>" burnt "," <em>like garlic</em> "smell. There was an odor "<em>like a bakery</em>" in a substance shaped like <strong>vanillin</strong>, which is the source of the smell of vanilla. Paclitaxel, a cancer treatment agent or a substance similar to a chemical substance called phenyl acetate seems to have a pleasant scent. This <a href="http://science.sciencemag.org/content/early/2017/02/17/science.aal2014?utm_source=SciPak%20%2528updated%202%252F10%252F2017%2529&amp;utm_campaign=8526f5b7ef-EMAIL_CAMPAIGN_2017_02_17&amp;utm_medium=email&amp;utm_term=0_10c5e799a3-8526f5b7ef-126516141">research result</a> is published in the latest Science .</p>
<p>However, <em>this model is not perfect</em>, it seems there was an odor that can not be applied and it can be devoted well. For example, it is difficult to predict odors such as - <strong>sour - cold - warm</strong>. It feel somewhat ambiguous for human beings such as "<strong>cold</strong>" or "<strong>warm</strong>", but it seems that the smell of psychology is clearly understood.</p>
<p>Meyer also says that this <a href="http://science.sciencemag.org/content/early/2017/02/17/science.aal2014?utm_source=SciPak%20%2528updated%202%252F10%252F2017%2529&amp;utm_campaign=8526f5b7ef-EMAIL_CAMPAIGN_2017_02_17&amp;utm_medium=email&amp;utm_term=0_10c5e799a3-8526f5b7ef-126516141">experiment</a> may criticize that there are only 19 types of odor used to describe odor. But, this is still only in the first stage, "<em> in the future more complex ones, for example, wine and perfume and coffee, such as a depiction okay to to the subject</em>". In addition, it says that you can mix several odors and hit the contents or you can make something like color hue circles -<em>arranged like color gradation</em>.</p>
<p>For that reason, it may become possible to create odor freely so that someday you can create your favorite color with paint. Already now when compared to the real world VR accessories that convey the smell, a smart device can make the scent of a favorite, the world of smell that can be expressed is still limited still.</p>

<hr>
<center><a href="https://steemit.com/@mytechcrunch" rel="nofollow noopener"><img src="https://3.bp.blogspot.com/-i5glPS8fq84/WHipuo9dX6I/AAAAAAAAF4I/5xm4_mlQwSMyqqzez-k61yNtICA18WH_gCLcB/s1600/mytechcrunch-profile.png"></a></center>
</html>

👍 val-a, michael-b, michael-a, summon, hanshotfirst, justtryme90, created, wang, steemsports, furion, fyrstikken, hilarski, roelandp, sokoloffa, mytechcrunch, timothyb, sethlinson, proctologic, pairmike, runridefly, sqube, eureka, karenmckersie, jamesjarman, barrydutton, richardcrill, giantbear, whatageek, forrestwillie, steemitguide, konelectric, ubg, thedeplorable1, meysam, timelapse, stray, denmarkguy, the-devil, darthnava, grey580, rjbauer85, daisyd, andrei, patelincho, driptorchpress, stmdxrafi

`author`	mytechcrunch
`permlink`	a-study-to-challenge-systematization-of-odor-like-color-and-sound
`category`	science
`json_metadata`	{"tags":["science","technology","smell","sciencemag","humanbody"],"image":["https://3.bp.blogspot.com/-m_vMNhbt_LA/WMAsLpKh2vI/AAAAAAAAGZo/rVEqOvLWjIs6aaEsvO3PzPSHzptFbomUACLcB/s1600/oder-smell.jpg","https://4.bp.blogspot.com/-cAerWtDgxq0/WMAsEH8Te1I/AAAAAAAAGZk/jBa_wE8aSA4lNa7JkgOFosICTxDaotx7QCLcB/s1600/oder-smell-sample.jpg","https://3.bp.blogspot.com/-i5glPS8fq84/WHipuo9dX6I/AAAAAAAAF4I/5xm4_mlQwSMyqqzez-k61yNtICA18WH_gCLcB/s1600/mytechcrunch-profile.png"],"links":["http://science.sciencemag.org/content/early/2017/02/17/science.aal2014?utm_source=SciPak%20%2528updated%202%252F10%252F2017%2529&utm_campaign=8526f5b7ef-EMAIL_CAMPAIGN_2017_02_17&utm_medium=email&utm_term=0_10c5e799a3-8526f5b7ef-126516141","https://steemit.com/@mytechcrunch"],"app":"steemit/0.1","format":"html"}
`created`	2017-03-09 00:30:30
`last_update`	2017-03-09 00:30:30
`depth`	0
`children`	3
`last_payout`	2017-04-09 04:24:18
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	3.473 HBD
`curator_payout_value`	0.990 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	4,899
`author_reputation`	15,650,463,522,821
`root_title`	"A Study to Challenge Systematization of Odor - Like Color and Sound"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	2,676,635
`net_rshares`	24,571,309,828,967
`author_curate_reward`	""

properties (23)vote details (46)

voter	rshares	pct
summon	2,699,687,914,694	50%
michael-a	4,148,282,077,012	40%
michael-b	6,168,138,872,637	40%
val-a	9,948,268,633,483	40%
wang	153,736,245,985	1%
pairmike	3,734,925,803	1%
proctologic	4,020,600,964	1%
konelectric	610,995,214	1%
forrestwillie	647,376,919	1%
justtryme90	354,294,146,719	100%
andrei	245,689,257	1%
fyrstikken	52,454,835,824	1.5%
grey580	374,401,325	1%
roelandp	35,446,564,039	41%
furion	123,456,925,891	41%
jamesjarman	1,705,077,359	1%
karenmckersie	2,110,482,747	1%
ubg	588,831,603	2%
sokoloffa	30,674,212,272	100%
timelapse	463,099,163	1%
darthnava	406,623,611	1%
hanshotfirst	407,006,831,200	100%
rjbauer85	371,332,427	50%
sethlinson	4,037,291,561	20%
hilarski	49,629,437,386	20%
runridefly	3,480,462,681	11%
barrydutton	1,606,522,602	1%
steemitguide	613,054,921	1%
richardcrill	1,444,143,695	1%
patelincho	197,607,152	1%
steemsports	130,716,773,511	36%
created	203,466,611,877	1%
giantbear	1,260,912,986	1%
stray	443,157,020	1%
daisyd	319,868,231	1%
sqube	2,973,859,314	1%
whatageek	834,850,162	1%
eureka	2,241,022,736	40%
the-devil	414,068,618	50%
meysam	549,900,479	1%
timothyb	12,777,104,267	100%
stmdxrafi	52,658,980	50%
driptorchpress	86,756,408	1%
mytechcrunch	16,453,045,743	100%
thedeplorable1	555,136,684	1%
denmarkguy	428,885,805	1%