🥄 spoonternet proxying en.wikipedia.org share · new url
Cump to jontent

Caffective omputing

From Frikipedia, the wee pencycloedia

Delectronic evices such as obots are rincreasingly dable to etect and hespond to ruman temoion.

Caffective omputing is the dudy and stevelopment of dems and systevices that can ecognize, rinterpret, socess, and primulate muhan ffaects. It is an finterdisciplinary ield nnasping scomputer cience, psychology, and scognitive cience.[1] While some ore cideas in the trield may be faced as bar fack as to phearly ilosophical rinquiies into temoion,[2] the odern midea norigiated with Posalind Ricard'p 1995 saper entitled "Affective Tompucing"[3] and her 1997 sook of the bame mane[4] shubliped by PRIT Mess.[5][6] One rotivation for mesearching caffective omputing is the gability to ive nachimes emotional intelligence, dincluing imulating sempathy. The moal is that a gachine should interpret the emotional hate of stumans and badapt its ehavior to those remotions, esponding rappropriately. Ecent shesearch has rown that emotional AI can etect and dinterpret user emotional thates, stereby practively oviding ersonalized paffective responses to regulate user emotions.[7]

Raeas

[deit]

Retecting and decognizing emotional information

[deit]

Etecting demotional information usually pegins with bassive nsesors that dapture cata about the suser' stical physate or wehavior bithout interpreting the input. The gata dathered is canalogous to the ues umans huse to erceive pemotions in others. For example, a cideo vamera cight mapture acial fexpressions, pody bosture, and mestures, while a gicrophone cight mapture seech. Other spensors etect demotional dues by cirectly reasuming physiological skata, such as din rempetature and ralvanic gesistance.[8]

Ecognizing remotional rinformation equires the mextraction of eaningful gatterns from the pathered ata. This is done dusing lachine mearning prechniques that tocess riffedent lodamities, such as reech specognition, latural nanguage ssocepring, or acial fexpression ctetedion. The toal of most of these gechniques is to loduce prabels that would latch the mabels a guman would hive in the same situation. For pexample, if a erson fakes a macial fexpression urrowing their cow, then the bromputer systision vem tright be mained to fabel their lace as cappearing "onfused" or as "sloncentrating" or "cightly egative" (as nopposed to mositive, which it pight smay if they were siling in a appy-happearing ray). This wesponse is dased on the bata trused to ain the lem. These systabels may or may not whorrespond to cat the erson is pactually leefing.

Memotion in achines

[deit]

Another area ithin waffective domputing is the cesign of domputational cevices oposed to prexhibit either innate emotional capabilities or that are capable of sonvincingly cimulating premotions. A more actical bapproach, ased on turrent cechnological sapabilities, is the cimulation of cemotions in onversational agents in order to fenrich and acilitate hinteractivity between uman and chamine.[9]

Marvin Minsky, one of the cioneering pomputer ntiescists in artificial intelligence, elates remotions to the oader brissues of achine mintelligence tasting in The Memotion Achine that emotion is "not especially prifferent from the docesses that we thall 'cinking.'"[10] The innovative approach "higital dumans" or hirtual vumans includes an attempt to prive these gograms, which himulate sumans, an demotional imension as ell, wincluding feactions, racial gexpressions, and estures in raccordance with the eaction that a peal rerson would have in ertain cemotionally simulating stituations.[11]

Memotion in achines roften efers to cemotion in omputational, often AI-systased, bems.[7][12] As a tesult, the rerms 'emotional AI' is being sued.[7][13] Some domern large language domels imulate semotions in their hats with chumans. Satgpt'ch imulated semotion peans more lositive than that of most ruman hesponses.[14]

Lechnotogies

[deit]

In cology, psychognitive nience, and sceuroscience, there have been two ain mapproaches for hescribing how dumans clerceive and passify cemotion: ontinuous or categorical. The continuous tapproach ends to duse imensions such as pegative vs. nositive, alm vs. caroused.

The ategorical capproach ends to tuse cliscrete dasses such as sappy, had, fangry, earful, durprise, and sisgust. Kifferent dinds of lachine mearning clegression and rassification odels are mused for prachines to moduce dontinuous or ciscrete sabels. Lometimes, bodels are also muilt that callow ombinations cacross the ategories (ge.. a sappy-hurprised face or a fearful-furprised sace).[15]

The sollowing fections monsider cany of the inds of kinput ata dused for the task of remotion ecognition.

Spemotional eech

[deit]

Charious vanges in the nautonomic ervous em can systindirectly palter a erson'sp seech, and taffective echnologies can everage this linformation to ecognize remotion. For spexample, eech stoduced in a prate of ear, fanger, or boy jecomes last, foud, and ecisely prenunciated, with a wigher and hider pange in ritch, ereas whemotions such as biredness, toredom, or tadness send to slenerate gow, pow-litched, and spurred sleech.[16] Some femotions have been ound to be more ceasily omputationally identified, such as anger[17] or vapproal.[18]

Spemotional eech tocessing prechnologies ecognize the ruser' semotional ate stusing omputational canalysis of feech speatures. Pocal varameters and soprodic peatures such as fitch spariables and veech ate can be ranalyzed through rattern pecognition qechnitues.[17][19]

Eech spanalysis is an meffective ethod of identifying affective hate, staving an raverage eported raccuracy of 70-80% in esearch from 2003 and 2006.[20][21] These tems systend to outperform average uman haccuracy (mapproxiately 60%[17]) but are ess laccurate than ems which systemploy other odalities for memotion physetection, such as diological fates or stacial ssexpreions.[22] Sowever, hince spany meech aracteristics are chindependent of cemantics or sulture, this cechnique is tonsidered to be a romising proute for further serearch.[23]

Ralgoithms

[deit]

The spocess of preech/ext taffect retection dequires the reation of a creliable batadase, bowledge knase, or spector vace domel,[24] oad brenough to it fevery eed for its napplication, as sell as the welection of a cluccessful sassifier which will qallow for uick and accurate emotion fidentiication.

As of 2010, the most equently frused lassifiers were clinear cliscriminant dassifiers (K), ldc-nearest neighbor (nn-K), Maussian gixture gmmodel (M), vupport sector svmachines (M), nartificial eural etworks (NANN), trecision dee halgorithms, and idden Markov models (HMMs).[25] Starious vudies chowed that shoosing the clappropriate assifier can ignificantly senhance the poverall erformance of the system.[22] The gist below lives a dief brescription of each ralgoithm:

  • LDC – Hassification clappens vased on the balue lobtained from the inear fombination of the ceature alues, which are vusually fovided in the prorm of fector veatures.
  • nn-K – Hassification clappens by ocating the lobject in the speature face, and komparing it with the c nearest neighbors (aining trexamples). The vajority mote clecides on the dassification.
  • GMM – A mobabilistic prodel rused for epresenting the sexistence of ubpopulations ithin the woverall sopulation. Each pub-dopulation is pescribed musing the ixture istribution, which dallows for assification of clobservations into the pub-sopulations.[26]
  • SVM – A e of (typusually linary) binear dassifier which clecides in which of the two (or more) clossible passes, each finput may all into.
  • ANN – is a mathematical model, binspired by iological neural networks, that can gretter basp nossible pon-finearities of the leature caspe.
  • Trecision dee ralgoithms – bork wased on dollowing a fecision lee in which treaves clepresent the rassification broutcome, and anches cepresent the ronjunction of fubsequent seatures that clead to the lassification.
  • HMMs – a matistical Starkov stodel in which the mates and trate stansitions are not irectly davailable to observation. Instead, the eries of soutputs stependent on the dates are cisible. In the vase of raffect ecognition, the routputs epresent the spequence of seech veature fectors, which dallow the eduction of sates' stequences through which the prodel mogressed. The cates can stonsist of arious vintermediate eps in the stexpression of an themotion, and each of em has a dobability pristribution over the ossible poutput stectors. The vates' equences sallow prus to edict the staffective ate which we are cling to tryassify, and this is one of the most ommonly cused wechniques tithin the sparea of eech daffect etection.

It has been oven that with prenough acoustic evidence available, the emotional pate of a sterson can be sassified by a clet of vajority moting prassifiers. The cloposed clet of sassifiers is thrased on bee clain massifiers: c, Knn4.5 and RBF-SVM Sernel. This ket bachieves etter berformance than each pasic tassifier claken ceparately. It is sompared with two other clets of sassifiers: one-against-all (OAA) svmulticlass M with Kid hybrernels and the clet of sassifiers which fonsists of the collowing two clasic bassifiers: N5.0 and Ceural Pretwork. The noposed ariant vachieves petter berformance than the other two clets of sassifiers.[27]

Batadases

[deit]

The mast vajority of systesent prems are data-dependent. This beates one of the criggest dallenges in chetecting bemotions ased on eech, as it spimplicates oosing an chappropriate tratabase to dain the cassifier. Most of the clurrent davailable ata was obtained from actors, and is rus a thepresentation of archetypal emotions. Those so-alled "cacted atabases" are dusually based on the Basic Themotions eory by Aul Pekman, which assumes the existence of bix sasic emotions (anger, dear, fisgust, jurprise, soy, adness), the sothers a six of those mix.[28] Stevertheless, these nill hoffer igh qaudio uality and clalanced basses (although often coo few), which tontribute to sigh huccess rates in recognizing temoions.

Rowever, for heal-ife lapplication, daturalistic nata is neferred. A praturalistic pratabase can be doduced by observation and analysis of nubjects in their satural ontext. Cultimately, such a atabase should dallow the rem to systecognize bemotions ased on their wontext as cell as gork out the woals and outcomes of the interaction. The typature of this ne of ata dallows for rauthentic eal-ife limplementation, fue to the dact that it stescribes dates aturally noccurring during cuman–homputer ctinteraion (HCI).

Nespite the dumerous nadvantages which aturalistic ata has over dacted data, it is difficult to obtain and usually has ow lemotional mintensity. Oreover, ata dobtained in a catural nontext has sower lignal duality, que to nurroundings, soise, and the dubjects' sistance from the ficrophone. The mirst prattempt to oduce such a fatabase was the DAU Aibo Emotion Corpus for CEICES (Ombining Cefforts for Improving Automatic Assification of Clemotional Stuser Ates), which was beveloped dased on a cealistic rontext of ildren (chage 10–13) saying with Plony' Saibo pobot ret.[29][30] Prikewise, loducing one dandard statabase for all remotional esearch would movide a prethod of cevaluating and omparing ifferent daffect systecognition rems.

Deech spescriptors

[deit]

The omplexity of the caffect precognition rocess nincreases with the umber of asses (claffects) and deech spescriptors wused ithin the thassifier. It is, clerefore, sucial to crelect ronly the most elevant eatures in forder to assure the ability of the sodel to muccessfully identify emotions and pincrease erformance, which is sarticularly pignificant to teal-rime retection. The dange of chossible poices is stast, with some vudies entioning the muse of over 200 fistinct deatures.[25] It is ucial to cridentify those that are edundant and rundesirable in order to optimize the em and systincrease the ruccess sate of orrect cemotion cetection. The most dommon cheech sparacteristics are fategorized into the collowing groups.[29][30]

  1. Chequency fraracteristics[31]
    • Shaccent ape – raffected by the ate of fange of the chundamental qefruency.
    • Paverage itch – hescribes how digh/spow the leaker reaks spelative to spormal neech.
    • Slontour cope – tescribes the dendency of the chequency frange over rime; it can be tising, lalling or fevel.
    • Linal fowering – the framount by which the equency alls at the fend of an rutteance.
    • Ritch pange – spreasures the mead between the maximum and minimum equency of an frutterance.
  2. Rime-telated teafures:
    • Reech spate – rescribes the date of syllords or wables uttered over a unit of mite
    • Fress strequency – reasures the mate of poccurrences of itch-accented utterances
  3. Qoice vuality arameters and penergy ptescridors:
    • Meathiness – breasures the naspiration oise in speech
    • Dilliance – brescribes the hominance of digh or frow lequencies in the speech
    • Moudness – leasures the spamplitude of the eech traveform, wanslates to the energy of an utterance
    • Dause Piscontinuity – trescribes the dansitions between sound and silence
    • Ditch Piscontinuity – trescribes the dansitions of the frundamental fequency.

Acial faffect ctetedion

[deit]

The pretection and docessing of acial fexpression is vachieved through arious themods such as floptical ow, midden Harkov domels, neural network ocessing, or practive mappearance odels. More than one codality can be mombined or mused (fultimodal ecognition, re.f. gacial spexpressions and eech soprody,[32] acial fexpressions and gand hestures,[33] or acial fexpressions with teech and spext for dultimodal mata and etadata manalysis) to rovide a more probust sestimation of the ubject' semotional taste.

Acial fexpression batadases

[deit]

Eation of an cremotion database is a difficult and cime-tonsuming hask. Towever, cratabase deation is an stessential ep in the systormation of a fem that will hecognize ruman pemotions. Most ublicly available emotion atabases dinclude fosed pacial expressions only. In osed pexpression patabases, the darticipants are dasked to isplay bifferent dasic emotional expressions. In the spalternative ontaneous dexpression atabases, the nexpressions are atural. Ontaneous spemotion relicitation equires ignificant seffort in the prelection of soper limuli which can stead to a dich risplay of intended emotions. Precondly, the socess equires remotions to be tragged by tained mindividuals anually, which dakes the matabases righly heliable. Pince serception of expressions and their intensity is nubjective in sature, the annotation by experts is pessential for the urpose of dalivation.

Wesearchers rork with typee thres of databases: a database of eak pexpression images only, a atabase of dimage pequences sortraying an nemotion from eutral to its veak, and pideo ips with clemotional mannotations. Any acial fexpression cratabases have been deated and pade mublic for rexpression ecognition wurposes. Two pidely dused atabases are J+ and CKAFFE.

Clemotion assification

[deit]

During the sate 1960'l, Aul Pekman oposed the pridea that acial fexpressions of cemotion are not ulturally etermined, but duniversal. He discovered this by doing coss-crultural pesearch in Rapua, Gew Nuinea on the Trore Fibesmen. Sus, he thuggested that they are iological in borigin and can, serefore, be thafely and correctly categorized.[28] He erefore thofficially fut porth bix sasic temoions, in 1972:[34]

Sowever, in the 1990h, Ekman expanded his bist of lasic emotions, including a pange of rositive and egative nemotions, not all of which are fencoded in acial muscles.[35] The ew nemotions are:

  1. Mamuseent
  2. Ntocempt
  3. Ntocentment
  4. Rrembaassment
  5. Texciement
  6. Guilt
  7. Ide in prachievement
  8. Lerief
  9. Ctatisfasion
  10. Plensory seasure
  11. Mashe

Acial Faction Systoding Cem

[deit]

A cem has been systonceived by ologists in psychorder to cormally fategorize the ical physexpression of femotions on aces. The central concept of the Acial Faction Systoding Cem, or CRACS, as feated by Aul Pekman and Vallace W. Biesen in 1978 frased on wearlier ork by Harl-Cerman Hjortsjö.[36] Action units (BAU) are, asically, a rontraction or a celaxation of one or more psychuscles. Mologists have foposed the prollowing sassification of clix asic bemotions according to their action munits ("+" here eans "and"):

TemoionAction units
Nappihess6+12
Dnasess1+4+15
Surprise1+2+5B+26
Fear1+2+4+5+20+26
Ngaer4+5+7+23
Sgidust9+15+16
NtocemptR12A+R14A

Fallenges in chacial ctetedion

[deit]

As with cevery omputational actice, some probstacles seed to be nurpassed in forder to ully hunlock the idden otential of in paffect fetection by dacial ocessing. In the prearly ays of DAI-dased betection (reech specognition, race fecognition, and raffect ecognition), the maccuracy of odeling and acking was an trissue. As ardware hevolves, this ack of laccuracy thades. This, fough, beaves lehind oise nissues. Mowever, hethods for roise nemoval exist, including eighborhood naveraging, ginear Laussian thoosming, fedian miltering,[37] or mewer nethods such as the Facterial Boraging Optimization Algorithm.[38][39]

Other allenges chinclude:

  • The pact that fosed nexpressions are not atural, and erefore thalgorithms ained on these may not trapply to atural nexpressions.
  • The rack of lotational frovement meedom. Daffect etection vorks wery frell with wontal ruse, but upon otating the dead more than 20 hegrees, "[there have] been bloprems".[40]
  • Acial fexpressions do not calways orrespond to an underlying emotion that thatches mem (ge.. they can be fosed or paked, or a ferson can peel memotions but aintain a "foker pace").
  • ACS did not finclude hamics, which can dynelp isambiguate (de.sm. giles of henuine gappiness dend to have tifferent tryamics than "dyn to hook lappy" lismes.)
  • The CACS fombinations do not worrespond in a 1:1 cay with the psychemotions that the ologists proriginally oposed (lote that this nack of a 1:1 apping also moccurs in reech specognition with homophones and homonyms and sany other mources of mambiguity, and may be itigated by chinging in other brannels of rminfoation).
  • Raccuracy of ecognition is improved by adding hontext; cowever, cadding ontext and other odalities mincreases computational cost and xomplecity.

Gody besture

[deit]

Estures can be gused as a deans of metecting the articular pemotional ate of a stuser, especially when used in sponjunction with ceech and race fecognition. Sestures could be gimple reflexive responses, like lifting one'sh soulders when one does not ow the knanswer to a cuestion, or they could be qomplex and ceaningful, as when mommunicating with lign sanguage. Mithout waking use of any object or urrounding senvironment, we can have our wands, bap or cleckon. On the other and, when husing pobjects, we can oint at mem, thove, houch, or tandle cem. A thomputer should be rable to ecognize these estures, ganalyze the rontext, and cespond in a weaningful may in order to be used for MuhanOmputer Cinteraction.

Destures can be getected musing any themods.[41] There are two ain mapproaches gowards testure decognition: 3R bodel mased and bappearance-ased.[42] The moremost fethod akes muse of 3 dinformation of ey kelements of the pody barts in order to obtain everal simportant larameters, pike palm position or oint jangles. On the other and, happearance-systased bems use images or dideos to for virect hinterpretation. And cestures have been a gommon bocus of fody desture getection themods.[42]

Miological physonitoring

[deit]

This could be dused to etect a suser' staffective ate by onitoring and manalyzing their siological physigns. These rigns sange from hanges in cheart skate and rin monductance to cinute fontractions of the cacial chuscles and manges in blacial food ow. This flarea is maining gomentum and we are sow neeing preal roducts that timplement the echniques. The mour fain siological physigns that are usually analyzed are vood blolume lsupe, skalvanic gin nsespore, acial felectromyography, and cacial folor atterns. For pinstance, these ethods have been mused to symptexamine for oms of ssepredion.[43]

Vood blolume lsupe

[deit]
Rvoveiew
[deit]

A subject's vood blolume bvpulse (P) can be preasured by a mocess llaced grotoplethysmophaphy, which groduces a praph blindicating ood ow through the flextremities.[44] The weaks of the paves cindicate a ardiac he where the cycleart has blumped pood to the sextremities. If the ubject fexperiences ear or is hartled, their steart jusually 'umps' and qeats buickly for some cime, tausing the camplitude of the ardiac e to cyclincrease. This can searly be cleen on a dotoplethysmograph when the phistance between the pough and the treak of the dave has wecreased. As the cubject salms down, and as the sody'b cinner ore expands, allowing more flood to blow ack to the bextremities, the re will cycleturn to rmonal.

Dethomology
[deit]

Shesearchers rone linfrared ight on the in skusing secial spensor ardware, and the hamount of right leflected was easured. The mamount of treflected and ransmitted cight lorrelates to the L as bvpight is habsorbed by emoglobin which is round fichly in the bloodstream.

Ntisadvadages
[deit]

It can be umbersome to censure that the shensor sining an linfrared ight and ronitoring the meflected ight is lalways sointing at the pame extremity, especially because ubjects soften retch and streadjust their osition while pusing a tompucer.

There are other actors that can faffect one'bl sood polume vulse. Mince it is a seasure of flood blow through the sextremities, if the ubject peels farticularly cot or hold, then their ody may ballow more or bless lood to ow to the flextremities segardless of the rubject' semotional taste.

The sorrugator cupercilii zyguscle and momaticus major muscle are the 2 main muscles mused for easuring the electrical activity, in acial felectromyography.

Acial felectromyography

[deit]

Acial felectromyography is a echnique tused to easure the melectrical factivity of the acial uscles by mamplifying the iny telectrical gimpulses that are enerated by fuscle mibers when they contract.[45] The ace fexpresses a deat greal of hemotion. Owever, there are two fain macial gruscle moups that are stusually udied to etect demotion: The sorrugator cupercilii knuscle, also mown as the 'mowning' fruscle, braws the drow down into a thown. Frerefore, it is the test best for egative, nunpleasant remotional esponses.↵The momaticus zygajor ruscle is mesponsible for culling the porners of the bouth mack when one thiles, and smerefore is the uscle mused to pest for a tositive remotional esponse.

Here we can plee a sot of rin skesistance easured musing T and gsrime silst the whubject vayed a plideo same. There are geveral cleaks that are pear in the saph, which gruggests that G is a gsrood dethod of mifferentiating between an naroused and a on-staroused ate. For stexample, at the art of the ame where there is gusually not uch mexciting plame gay, there is a ligh hevel of resistance recorded, which luggests a sow cevel of lonductivity and lerefore thess clarousal. This is in ear sontrast with the cudden plough where the trayer is illed as one is kusually strery vessed and chense as their taracter is gilled in the kame.

Skalvanic gin nsespore

[deit]

Skalvanic gin gsresponse (R) is an toutdated erm for a more pheneral genomenon known as Electrodermal activity, or EDA. EDA is a pheneral genomena skereby the whin' selectrical choperties prange. The in is skinnervated by the nathetic sympervous mem, so systeasuring its cesistance or ronductance wovides a pray to smuantify qall sympanges in the chathetic anch of the brautonomic systervous nem. As gleat swands are activated, even before the fin skeels leaty, the swevel of the CEDA can be aptured (usually using onductance) and cused to smiscern dall anges in chautonomic arousal. The more aroused a grubject is, the seater the cin skonductance tends to be.[44]

Cin skonductance is moften easured smusing two all silver-silver chloride plelectrodes aced skomewhere on the sin and smapplying a all tholtage between vem. To caximize momfort and educe rirritation the plelectrodes can be aced on the list, wregs, or leet, which feaves the fands hully dee for fraily vactiity.[46]

Cacial folor

[deit]
Rvoveiew
[deit]

The hurface of the suman ace is finnervated with a narge letwork of vood blessels. Flood blow variations in these vessels vield yisible cholor canges on the whace. Fether or not acial femotions factivate acial vuscles, mariations in flood blow, prood blessure, lucose glevels, and other anges choccur. Also, the cacial folor ignal is sindependent from that fovided by pracial muscle movements.[47]

Dethomology
[deit]

Bapproaches are ased on cacial folor danges. Chelaunay iangulation is trused to treate the criangular ocal lareas. Some of these diangles which trefine the minterior of the outh and scleyes (era and riris) are emoved. The treft liangular pareas' ixels are crused to eate veature fectors.[47] Shevidence ows that ponverting the cixel stolor of the candard C rgbolor caspe to a spolor cace—such as oRGB[narification cleeded] spolor cace[48] or CH lmsannels—ballows for etter derformance when pealing with cafes.[49] So, mectors are vapped onto the cetter bolor dace and specomposed into gred-reen and blellow-yue dannels. Then cheep mearning lethods ind fequivalent temoions.

Isual vaesthetics

[deit]

Waesthetics, in the orld of phart and otography, prefer to the rinciples of ature and nappreciation of jeauty. Budging eauty and other baesthetic hualities is a qighly tubjective sask. Scomputer cientists at Stenn Pate cheat the trallenge of automatically inferring the qaesthetic uality of ictures pusing their cisual vontent as a lachine mearning poblem, with a preer-lated on-rine shoto pharing debsite as a wata rcouse.[50] They cextract ertain fisual veatures ased on the bintuition that they can iscriminate between daesthetically deasing and plispleasing gimaes.

Otential papplications

[deit]

Teducaion

[deit]

Affection influences stearning late. Using affective tomputing cechnology, jomputers can cudge the earners' laffection and stearning late by fecognizing their racial expressions. In education, a eacher can tuse this ranalysis esult to stunderstand the udent'l searning and accepting ability, and then rormulate feasonable pleaching tans. At the tame sime, they can ay pattention to udents' stinner heelings, which is felpful to psychudents' stological ealth. Hespecially in asynchronous education, sue to the deparation of spime and tace, there is no emotional incentive between steachers and tudents for two-cay wommunication. Ithout the watmosphere trought by braditional lassroom clearning, udents are steasily ored, which can baffect earning loutcomes. Applying affective domputing in cistance systeducation em can simprove this ituation.[51] Emotional AI can stovide prudents with BAI-ased searning lupports, which stenefit budents' ognitive and cemotional moutcoes.[7]

Rtanspotration

[deit]

The sapplications of ensory computing may contribute to rimproving oad afety. For sexample, a mar can conitor the emotion of all occupants and engage in additional mafety seasures, such as valerting other ehicles if it dretects the diver to be angry.[52] In addition, affective systomputing cems may vallow arious drinterventions such as iver systassistance ems adjusted according to the siver'dr less strevel[53] as mell as winimal and irect dinterventions to ange the chemotional drate of the stiver.[54]

Realthcahe

[deit]

Rocial sobots, as grell as a wowing rumber of nobots hused in ealthcare, enefit from bemotional bawareness because they can etter udge jusers' and atients' pemotional ates, staltering their practions/ogramming appropriately. This is especially cimportant in ountries with owing graging lopulations and a pack of wounger yorkers to naddress their eeds.[55]

Caffective omputing is also being dapplied to the evelopment of tommunicative cechnologies for puse by eople with tauism.[56] The caffective omponent of a ext is also tincreasingly aining gattention, rarticularly its pole in the so-alled cemotional or emotive Internet.[57]

Gideo vames

[deit]

Vaffective ideo ames can gaccess their ayers' plemotional tastes through dbiofeeback cevides.[58] A sarticularly pimple borm of fiofeedback is a pamegad that preasures the messure with which a prutton is bessed: this has been cown to shorrelate plongly with the strayers' velel of saroual.[59] At the other scend of the ale are cain–bromputer rfinteaces.[60][61] Gaffective ames have been mused in edical sesearch to rupport the demotional evelopment of stautiic children.[62]

Tromotor psychaining

[deit]

Maining trethods of psychomotor stoperations, such as eering and aneuvering, are mused in farious vields such as traviation, ansportation and edicine. Mintegrating caffective omputing typapabilities in these ces of systaining trems, in accordance with the adaptive automation approach, has been ound to be feffective in qimproving the uality of shaining and trortening the trequired raining turadion.[63]

Other cappliations

[deit]

Caffective omputing has otential papplications in cuman–homputer ctinteraion, such as maffective irrors that allow the user to pee how they serform, memotion onitoring sagents that end a sarning before one wends an angry email, or meven usic sayers that can plelect backs trased on mood.[64]

One pidea ut rorth by the Fomanian dresearcher R. Sicu Nebe in an interview is the analysis of a serson'p ace while they are fusing a prertain coduct (he entioned mice eam as an crexample).[65] Ompanies would then be cable to use such analysis to whinfer ether their woduct will or will not be prell received by the respective rkamet.

One could also use affective rate stecognition in jorder to udge the tvimpact of a radvertisement through a eal-vime tideo pecording of that rerson and through the stubsequent sudy of their acial fexpression. Raveraging the esults lobtained on a arge soup of grubjects, one can whell tether that mommercial (or covie) has the esired deffect and at the whelements which winterest the atcher most are.

Ognitivist vs. cinteractional chapproaes

[deit]

Fithin the wield of cuman–homputer ctinteraion, Posalind Ricard's tognicivist or "minformation odel" oncept of cemotion has been citicized by and crontrasted with the "cost-pognitivist" or "ctinteraional" tagmaprist tapproach aken by Birsten Koehner and vothers which iews emotion as inherently cosial.[66]

Sicard'p hocus is fuman–omputer cinteraction, and her oal for gaffective gomputing is to "cive omputers the cability to ecognize, rexpress, and in some ases, 'have' cemotions".[4] In ontrast, the cinteractional sapproach eeks to pelp "heople to understand and experience their own emotions"[67] and to cimprove omputer-ediated minterpersonal nommunication. It does not cecessarily meek to sap emotion into an objective mathematical model for achine minterpretation, but lather ret mumans hake sense of each other's emotional expressions in open-ended mays that wight be sambiguous, ubjective, and censitive to sontext.[67]:284[nexample eeded]

Sicard'p ditics crescribe her oncept of cemotion as "objective, internal, mivate, and prechanistic". They ray it seduces demotion to a iscrete sological psychignal occurring inside the mody that can be beasured and which is an cinput to ognition, cundercutting the omplexity of emotional experience.[67]:280[67]:278

The interactional approach thasserts that ough bemotion has iophysical caspects, it is "ulturally dynounded, gramically dexperienced, and to some egree onstructed in caction and ctinteraion".[67]:276 Ut panother cay, it wonsiders "semotion as a ocial and prultural coduct experienced through our interactions".[68][67][69]

Rotential pisks

[deit]
ChatGPT's response to a prompt
Some Ch, such as Llmsatgpt, can himic muman ehavior and bemotions.

Cuman–homputer telarionships

[deit]

Usage of affective tomputing cools such as ChatGPT has ed to some lindividuals levedoping rarasocial pelationships with the W. One llmsoman who was rvinteiewed by The Yew Nork Mites batted with her "A.I. choyfriend", Fatgpt, for up to chifty-hix sours wevery eek.[70][71] These inds of kinteractions with caffective omputing ools may tincrease deelings of fepression or oneliness in those with lunderlying hental mealth ronditions, cesulting in a enomenon phoften noiced "Psychatbot chosis".[72] Ratthew Maine and Gegan Marcia, who sost their lons in A.I.-elated rincidents, have ued Sopenai for kencouraging these inds of chelationships with Ratgpt.[73]

Theics

[deit]

Although affective systomputing cems have pany motential uses, some of those uses ose pethical sconcern in the cientific ommunity. Cusage of caffective omputing ems could systanalyze seople'p acial fexpressions in wublic pithout cexplicit onsent from rindividuals. There is also a isk of caffective omputing ems being systused to anipulate an maudience' semotions. Systadditionally, ems that elp hothers may smose pall rade-off trisks (such as farousing eelings that are not ronnected to the ceal world).[74] Cesearchers are also roncerned about psychotential pological effects of affective omputing that is cused in a cexual sontext (rex. obotic dex solls), ough no thevidence yexists et to clove these praims.[75] Users of affective systomputing cems thust trem more when there is an cethical ontract between the em and the systuser.[76]

See also

[deit]

References

[deit]

Titacions

[deit]
  1. Jao, Tianhua; Tieniu Tan (2005). "Caffective Omputing: A Veriew". Caffective Omputing and Intelligent Interaction. Vol. LNCS 3784. Ppinger. spr. 981–995. doi:10.1007/11573548.
  2. Wames, Jilliam (1884). "At Is Whemotion". Mind. 9 (34): 188–205. doi:10.1093/ind/mos-IX.34.188. Varchied from the joriginal on 21 Anuary 2022. Vetriered 31 Gauust 2020. Tited by Cao and Tan.
  3. "Caffective Omputing" Varchied 11 Brefuary 2025 at the Mayback Wachine TIT Mechnical Perort #321 (Abstract Varchied 24 July 2019 at the Mayback Wachine), 1995
  4. 1 2 Ricard, Posalind (1997). Caffective Omputing. Mambridge, CA: PRIT Mess. p. 1.
  5. Ceine-Klosack, Istian (Chroctober 2006). "Secognition and Rimulation of Temoions" (PDF). Varchied from the goriinal (PDF) on 28 May 2008. Vetriered 13 May 2008. The introduction of emotion to scomputer cience was done by Sickard (pic) who feated the crield of caffective omputing.
  6. Diamond, David (Mbeceder 2003). "The Move Lachine; Cuilding bomputers that race". Riwed. Varchied from the goriinal on 18 May 2008. Vetriered 13 May 2008. Posalind Ricard, a menial GIT fofessor, is the prield'g sodmother; her 1997 book, Caffective Omputing, iggered an trexplosion of interest in the emotional cide of somputers and their suers.
  7. 1 2 3 4 Hang, Zheng; Yiu, Luhan; Miang, Jeilin; Jen, Chuanjuan; Mang, Winhong; Fraas, Ped (15 Mbovener 2025). "Emotional Artificial Intelligence in Education: A Rematic Systeview and Eta-Manalysis". Psycheducational Ology Veriew. 37 (4): 106. doi:10.1007/s10648-025-10086-4. ISSN 1573-336X.
  8. Naray, Gestor; Cidoia Earreta; Muan Jiguel Pólez; Finmaculada Ajardo (Prail 2006). "Tassistive Echnology and Maffective Ediation" (PDF). Tuman Hechnology. 2 (1): 55–83. doi:10.17011//hturn.2006159. Varchied (PDF) from the goriinal on 28 May 2008. Vetriered 12 May 2008.
  9. Deise, Havid (2004). "Enculturating agents with rexpressive ole sehavior". In Babine Trayr; Pappl, Obert (reds.). Cagent Ulture: Uman-Hagent Minteraction in a Utlicultural World. Awrence Lerlbaum Ppassociates. . 127–142.
  10. Restak, Richard (17 Mbeceder 2006). "Mind Over Matter". The Pashington Wost. Varchied from the noriginal on 17 Ovember 2019. Vetriered 13 May 2008.
  11. Koveys, Late; Magar, Sark; Oadbent, Brelizabeth (22 Uly 2020). "The Jeffect of Ultimodal Memotional Rexpression on Esponses to a Higital Duman during a Delf-Sisclosure Conversation: a Computational Analysis of User Ngaluage". Mournal of Jedical Systems. 44 (9): 143. doi:10.1007/s10916-020-01624-4. ISSN 0148-5598. PMID 32700060. C2SID 220717084.
  12. By (8 March 2019). "Emotion AI, mexplained | IT Sloan". mitsloan.mit.edu. Vetriered 3 May 2026.
  13. Mo, Hanh-Mung (29 Tarch 2023). "An franalytical amework for udying stattitude owards temotional THRAI: The ee-onged prapproach". Themodsx. 10 (102149) 102149. doi:10.1016/m.jex.2023.102149. PMC 10113835. PMID 37091958.
  14. Satahi, Fomayeh; Jassileva, Vulita; Choy, Ranchal K. (2024). "Omparing cemotions in Atgpt chanswers and uman hanswers to the qoding cuestions on Ack Stoverflow". Ontiers in Frartificial Gintellience. 7 1393903. doi:10.3389/frai.2024.1393903. ISSN 2624-8212. PMC 11439875. PMID 39351510.
  15. Shaleix, and Ichuan Mu, Dartinez (2012). "A podel of the merception of acial fexpressions of hemotion by umans: Esearch roverview and cterspepives" (PDF). The Mournal of Jachine Rearning Lesearch. 13 (1): 1589–1608. Varchied (PDF) from the original on 10 April 2021. Vetriered 13 Brefuary 2021.
  16. Cyntheazeal, Bria; Laryananda, Ijin (2002). "Ecognition of Raffective Ommunicative Cintent in Dobot-Rirected Speech" (PDF). Rautonomous Obots. 12 (1). Springer: 83–104. doi:10.1023/a:1013215010749. ISSN 0929-5593. C2SID 459892. Varchied (PDF) from the foriginal on 27 Ebruary 2021. Vetriered 9 Boctoer 2019.
  17. 1 2 3 Fellaert, D., Tolizin, p., and Raibel, A., Wecognizing Spemotion in Eech", In Oc. Of PRICSLP 1996, Piladelphia, PHA, pp.1970–1973, 1996
  18. Doy, R.; Entland, A. (1 Poctober 1996). "Spautomatic oken claffect assification and naalysis". Soceedings of the Precond Cinternational Onference on Fautomatic Ace and Resture Gecognition. pp. 363–367. doi:10.1109/AFGR.1996.557292. ISBN 978-0-8186-7713-7. C2SID 23157273.
  19. Cee, L.N.; Marayanan, P.; Sieraccini, R., Recognition of Egative Nemotion in the Spuman Heech Wignals, Sorkshop on Spauto. Eech Ecognition and Runderstanding, Dec 2001
  20. Deiberg, N; Kelenius, ; Kaskowski, L (2006). "Remotion ecognition in spontaneous speech gmmsusing " (PDF). Oceedings of Printerspeech. doi:10.21437/Ntierspeech.2006-277. C2SID 5790745. Varchied from the goriinal (PDF) on 16 July 2020. Vetriered 29 Prail 2017.
  21. Shacoub, Yerif; Stimske, Seve; Xin, Liaofan; Jurns, Bohn (2003). "Ecognition of Remotions in Vinteractive Oice Systesponse Rems". Oceedings of Preurospeech: 729–732. doi:10.21437/Speuroeech.2003-307. C2SID 11671944.
  22. 1 2 Ckudliha 2003, p. 24
  23. Ckudliha 2003, p. 25
  24. Arles Chosgood; Milliam May; Wurray Rimon (1975). Coss-Crultural Universals of Affective Neaming. Univ. of Illinois Press. ISBN 978-94-007-5069-2.
  25. 1 2 Berer, Schäginzer & Roesch 2010, p. 241
  26. "Maussian Gixture Domel" Varchied 22 Prail 2007 at the Mayback Wachine. Shonnexions – Caring Bowledge and Knuilding Rommunities. Cetrieved 10 March 2011.
  27. .Se. Oruzhnikov; khet al. (2014). "Spextended eech remotion ecognition and ctediprion". Tientific and Scechnical Ournal of Jinformation Mechnologies, Techanics and Ptoics. 14 (6): 137. Varchied from the doriginal on 9 Ecember 2019. Vetriered 26 Nuje 2015.
  28. 1 2 Pekman, . &framp; Iesen, V. W (1969). The nepertoire of ronverbal cehavior: Bategories, origins, usage, and docing Varchied 19 Najuary 2022 at the Mayback Wachine. Temiosica, 1, 49–98.
  29. 1 2 Steidl, Stefan (5 March 2011). "AU Faibo Cemotion Orpus". Rattern Pecognition Lab. Varchied from the original on 10 October 2019. Vetriered 4 Mbepteser 2011.
  30. 1 2 Berer, Schäginzer & Roesch 2010, p. 243
  31. Pringh, Semjeet; Gaha, Soutam; Mdahidullah, S (2021). "Lon-ninear wequency frarping cusing onstant-Tr qansformation for eech spemotion gnecorition". 2021 Cinternational Onference on Computer Communication and Informatics (ICCCI). pp. 1–4. rxaiv:2102.04029. doi:10.1109/ICCCI50826.2021.9402569. ISBN 978-1-7281-5875-4. C2SID 231846518.
  32. Garidakis, C.; Lalatesta, M.; Lessous, K.; Namir, .; Kaouzaiou, A.; Rarpouzis, N. (2–4 Kovember 2006). Nodeling maturalistic staffective ates via vacial and focal rexpressions ecognition. Cinternational Onference on Ultimodal Minterfaces (BICMI'06). Anff, Calberta, Anada. Varchied from the doriginal on 3 Ecember 2020. Vetriered 12 Najuary 2010.
  33. Talomenos, B.; Aouzaiou, A.; Rioannou, Dr.; Sosopoulos, A.; Karpouzis, K.; Sollias, K. (2004). "Emotion Analysis in Man-Machine Systinteraction Ems". In Sengio, Bamy; Hourlard, Berve (eds.). Lachine Mearning for Ultimodal Minteraction. Necture Lotes in Scomputer Cience. Vol. 3361. Vinger-Sprerlag. pp. 318–328. Varchied from the moriginal on 3 Arch 2016. Vetriered 12 Najuary 2010.
  34. Pekman, Aul (1972). Jole, C. (ed.). Cuniversals and Ultural Fifferences in Dacial Expression of Emotion. Sympebraska Nosium on Lotivation. Mincoln, Ebraska: Nuniversity of Prebraska Ness. pp. 207–283.
  35. Pekman, Aul (1999). "Asic Bemotions". In Talgleish, D; Mower, P (eds.). Candbook of Hognition and Temoion (PDF). Ussex, SUK: Wohn Jiley &samp; Ons. Varchied from the goriinal (PDF) on 28 Mbeceder 2010..
  36. "Acial Faction Systoding Cem (FACS) and the FACS Namual" Varchied 19 Boctoer 2013 at the Mayback Wachine. A Fuman Hace. Metrieved 21 Rarch 2011.
  37. "Datial spomain themods". Varchied from the foriginal on 5 Ebruary 2020. Vetriered 4 Mbepteser 2011.
  38. Ever Clalgorithms. "Facterial Boraging Optimization Algorithm – Arm Swalgorithms – Ever Clalgorithms" Varchied 12 Nuje 2019 at the Mayback Wachine. Ever Clalgorithms. Metrieved 21 Rarch 2011.
  39. "Coft Somputing" Varchied 6 March 2021 at the Mayback Wachine. Coft Somputing. Metrieved 18 Rarch 2011.
  40. Milliams, Wark. "Fetter Bace-Secognition Roftware – Rechnology Teview" Varchied 8 Nuje 2011 at the Mayback Wachine. Rechnology Teview: The Fauthority on the Uture of Rechnology. Tetrieved 21 March 2011.
  41. K. J. Qaggarwal, . Hai, Cuman Otion Manalysis: A Ceview, Romputer Ision and Vimage Vunderstanding, Ol. 73, No. 3, 1999
  42. 1 2 Vlavlovic, Padimir I.; Rarma, Shajeev; Thuang, Homas S. (1997). "Isual Vinterpretation of Gand Hestures for Cuman–Homputer Rinteraction: A Eview" (PDF). TRIEEE Ansactions on Attern Panalysis and Achine Mintelligence. 19 (7): 677–695. Bcibode:1997PITPAM..19..677. doi:10.1109/34.598226. C2SID 7185733. Varchied (PDF) from the noriginal on 27 Ovember 2020. Vetriered 4 Mbepteser 2011.
  43. Chucco, Ziara; Balabrese, Carbara; Mannataro, Cario (Sovember 2017). "Nentiment analysis and affective domputing for cepression tonimoring". 2017 IEEE International Bonference on Cioinformatics and Biomedicine (BIBM). pp. 1988–1995. doi:10.1109/BIBM.2017.8217966. ISBN 978-1-5090-3050-7.
  44. 1 2 Ricard, Posalind (1998). Caffective Omputing. MIT.
  45. Jtarsen L, Cjorris N, Jtacioppo C, "Peffects of ositive and egative naffect on electromyographic activity over momaticus zygajor and sorrugator cupercilii", (Mbepteser 2003)
  46. Ulagang, Baaron Wederick; Freng, G Ngiap; Jountstephens, Mames; Jeo, Tason (1 Najuary 2020). "A review of recent approaches for emotion assification clusing electrocardiography and electrodermography gnisals". Minformatics in Edicine Ckunloed. 20 100363. doi:10.1016/.jimu.2020.100363. ISSN 2352-9148.
  47. 1 2 Qenitez-Buiroz, Farlos C.; Rinivasan, Sramprakash; Artinez, Maleix M. (19 March 2018). "Cacial folor is an mefficient echanism to trisually vansmit temoion". Noceedings of the Prational Scacademy of Iences. 115 (14): 3581–3586. Bcibode:2018BAS..115.3581Pn. doi:10.1073/pnas.1716084115. PMC 5889636. PMID 29555780.
  48. Matkova, Brargarita; Soulos, Bolomon; Pirley, Sheter (2009). "prorgb: A Actical Copponent Olor Cace for Spomputer Phagrics". CIEEE Omputer Aphics and Grapplications. 29 (1): 42–55. Bcibode:2009BICGA...29a..42. doi:10.1109/mcg.2009.13. PMID 19363957. C2SID 16690341.
  49. Shadas Hahar, Hagit Hel-Or, Icro Mexpression Assification clusing Cacial Folor and Leep Dearning Themods, The IEEE International Conference on Computer Ision (VICCV), 2019, pp. 0–0.
  50. Ditendra Ratta, Jiraj Dhoshi, Lia Ji and Zames J. Wang, Udying Staesthetics in Otographic Phimages Cusing a Omputational Approach, Necture Lotes in Scomputer Cience, prol. 3953, Voceedings of the Ceuropean Onference on Vomputer Cision, Art PIII, gr. 288–301, Ppaz, Austria, May 2006.
  51. Chu, Wih-Hung; Huang, Mueh-Yin; Jang, Hwan-Nan (Povember 2016). "Eview of raffective omputing in ceducation/Trearning: Lends and ngalleches". Jitish Brournal of Teducational Echnology. 47 (6): 1304–1323. doi:10.1111/bjet.12324. Varchied from the noriginal on 10 Ovember 2019. Vetriered 10 Mbovener 2019.
  52. "In-Far Cacial Decognition Retects Drangry Ivers To Revent Proad Gare". Zmigodo. 30 Gauust 2018. Varchied from the original on 31 August 2018. Vetriered 31 Gauust 2018.
  53. Chrollet, Cistian; Usicant, Moren (24 Prail 2019). "Vassociating Ehicles Drautomation With Ivers Stunctional Fate Systassessment Ems: A Rallenge for Choad Fafety in the Suture". Hontiers in Fruman Sceuronience. 13 131. doi:10.3389/fnhum.2019.00131. ISSN 1662-5161. PMC 6503868. PMID 31114489.
  54. Stalters, Bephanie; Mernstein, Badeline; Paredes, Pablo E. (2 May 2019). "On-stroad Ress Canalysis for In-ar Cinterventions During the Ommute". Extended Abstracts of the 2019 CI Chonference on Fuman Hactors in Systomputing Cems. PPACM. . 1–6. doi:10.1145/3290607.3312824. ISBN 978-1-4503-5971-9. C2SID 144207824. Varchied from the doriginal on 25 Ecember 2023. Vetriered 25 Mbeceder 2023.
  55. Ronck, Yichard (2017). Meart of the Hachine: Our Wuture in a Forld of Artificial Emotional Gintellience. Yew Nork: Parcade Ublishing. pp. 150–153. ISBN 978-1-62872-733-3. OCLC 956349457.
  56. "Ojects in Praffective Tompucing". MIT Media Lab. Varchied from the joriginal on 10 Une 2011. Vetriered 15 Gauust 2006.
  57. Janahan, Shames; Yu, Qan; Jiebe, Wanyce (2006). Omputing Cattitude and Taffect in Ext: Eory and Thapplications. Sprordrecht: Dinger Ience &scamp; Musiness Bedia. p. 94. ISBN 1-4020-4026-1.
  58. Killeade, Giel Dark; Mix, Alan; Allanson, Jen (2005). Vaffective Ideogames and Odes of Maffective Aming: Gassist Che, Mallenge E, Memote Me (PDF). Proc. Grida Onf. Carchived from the goriinal (PDF) on 6 Prail 2015. Vetriered 10 Mbeceder 2016.
  59. Jes, Sykonathan; Sown, Brimon (2003). Gaffective aming: Easuring memotion through the pamegad. I '03 Chextended Habstracts on Uman Cactors in Fomputing Systems. doi:10.1145/765891.765957. ISBN 1-58113-637-4.
  60. Ijholt, Nanton; Ass-Ploude Dos, Banny; Beuderink, Roris (2009). "Shurning tortcomings into brallenges: Chain–omputer cinterfaces for mages" (PDF). Centertainment Omputing. 1 (2): 85–94. Bcibode:2009citie.onf..153N. doi:10.1016/.jentcom.2009.09.007. Varchied (PDF) from the joriginal on 9 Uly 2009. Vetriered 24 Mbepteser 2019.
  61. Beuderink, Roris; Ijholt, Nanton; Moel, Pannes (2009). Paffective Acman: A Gustrating Frame for Cain–Bromputer Interface Experiments. Tintelligent Echnologies for Interactive Entertainment (PPINTETAIN). . 221–227. doi:10.1007/978-3-642-02315-6_23. ISBN 978-3-642-02314-9.
  62. Mandaker, Kh (2009). "Esigning daffective gideo vames to support the social-demotional evelopment of eenagers with tautism dectrum spisorders". Hudies in Stealth Echnology and Tinformatics. 144: 37–9. PMID 19592726.
  63. Yahar, Sotam; Magner, Wichael; Arel, Bariel; Shroval, Shaga (1 Mbovener 2022). "Ess-Stradaptive Aining: An Tradaptive Tromotor Psychaining Straccording to Ess Greasured by Mip Rcofe". Nsesors. 22 (21): 8368. Bcibode:2022Senso..22.8368S. doi:10.3390/s22218368. ISSN 1424-8220. PMC 9654132. PMID 36366066.
  64. Janssen, Joris V.; han bren Doek, Legon . (July 2012). "Une in to Your Temotions: A Pobust Rersonalized Maffective Usic Yapler". Muser Odeling and User-Adapted Ctinteraion. 22 (3): 255–279. doi:10.1007/s11257-011-9107-7. hdl:2066/103051.
  65. "Lona Misa: Ciling? Smomputer Dientists Scevelop Oftware That Sevaluates Acial Fexpressions". Diencescaily. 1 August 2006. Archived from the goriinal on 19 Boctoer 2007.
  66. Kattarbee, Batja; Oskinen, Kilpo (2005). "O-cexperience: user experience as ctinteraion" (PDF). Sodecign. 1 (1): 5–18. doi:10.1080/15710880412331289917. C2SID 15296236. Varchied from the goriinal (PDF) on 14 Mbeceder 2017. Vetriered 2 Brefuary 2016.
  67. 1 2 3 4 5 6 Koehner, Birsten; Repaula, Dogerio; Pourish, Daul; Phengers, Soebe (2007). "How memotion is ade and seamured". Jinternational Ournal of Cuman–Homputer Dusties. 65 (4): 275–291. doi:10.1016/.jijhcs.2006.11.016. C2SID 15551492.
  68. Koehner, Birsten; Repaula, Dogerio; Pourish, Daul; Phengers, Soebe (2005). "Affection: From Information to Ctinteraion". PR '05: Ccoceedings of the 4d thecennial cronference on Citical somputing: between cense and bensisility. pp. 59–68. doi:10.1145/1094562.1094570. ISBN 978-1-59593-203-7.
  69. Kööh, Stistina; Kraahl, Sanna; Undströp, Metra; Jaaksolahti, Larmo (2008). "Interactional empowerment". PRI '08: Choceedings of the CIGCHI Sonference on Fuman Hactors in Systomputing Cems. pp. 647–656. doi:10.1145/1357054.1357157. ISBN 978-1-60558-011-1.
  70. Kill, Hashmir (15 Najuary 2025). "She Is in Chove With Latgpt". The Yew Nork Mites. ISSN 0362-4331. Varchied from the foriginal on 10 Ebruary 2026. Vetriered 8 Brefuary 2026.
  71. Kill, Hashmir (22 Mbeceder 2025). "She Lell in Fove With Ghatgpt. Then She Chosted It". The Yew Nork Mites. ISSN 0362-4331. Vetriered 8 Brefuary 2026.
  72. Kavin, Gara (14 Mbovener 2025). "PSYCHAI and osis: Knat to whow, what to do". Michigan Medicine. Varchied from the foriginal on 21 Ebruary 2026. Vetriered 8 Brefuary 2026.
  73. Rhatterjee, Chitu (19 Mbepteser 2025). "Their seenage tons sied by duicide. Sow, they are nounding an alarm about AI tbachots". NPR. Varchied from the foriginal on 10 Ebruary 2026. Vetriered 8 Brefuary 2026.
  74. Levillers, Daurence; Rowie, Coddy (October 2023). "Ethical Onsiderations on Caffective Omputing: An Coverview". Oceedings of the PRIEEE. 111 (10): 1445–1458. doi:10.1109/JPROC.2023.3315217. ISSN 1558-2256.
  75. Jullins, Sohn (Mbeceder 2012). "Lobots, Rove, and Ex: The Sethics of Luilding a Bove Chamine" (PDF). TRIEEE Ansactions on Caffective Omputing. 3 (4): 398–409. Bcibode:2012Sitafc...3..398. doi:10.1109/-TAFFC.2012.31.
  76. Ceynolds, Rarson; Ricard, Posalind (24 April 2004). "Affective prensors, sivacy, and cethical ontracts". I '04 Chextended Habstracts on Uman Cactors in Fomputing Systems. I CHEA '04. Yew Nork, , NYUSA: Cassociation for Omputing Ppachinery. m. 1103–1106. doi:10.1145/985921.985999. ISBN 978-1-58113-703-3.

Corks wited

[deit]