Te Whakamahi i te Ariā a Bayes i roto i te Tūponotanga

Te Whakamahi i te Ariā a Bayes i roto i te Tūponotanga

Ko te ariā a Bayes, i tapaina i muri i a Reverend Thomas Bayes, he ariā taketake i roto i te ao o te ariā tūponotanga. Mā tēnei ariā ka taea te whakahou i te tūponotanga o tētahi ariā i runga i ngā taunakitanga hou. Kua noho hei taputapu matua i roto i ngā momo mara, tae atu ki ngā tatauranga, te ako mīhini, te tautuhi mate hauora, me ngā tukanga whakatau. Ko te tikanga, ka taea te whakamahi i te ariā a Bayes hei whakahou i ngā matapae, i ngā ariā rānei kua tukuna he raraunga hou, he raraunga tāpiri rānei.

Te Mārama ki te Ariā a Bayes

Ko te ariā a Bayes he kīanga pāngarau ōkawa e hono ana i te tūponotanga o nāianei ki te tūponotanga o mua. E whai ake nei te whakapuakitanga:

\[ P(A|B) = \frac{P(B|A) \cdot P(A)}{P(B)} \]

Kei hea:
– Ko te \( P(A|B) \) te tūponotanga o muri, te tūponotanga rānei o te kaupapa A i te mea he pono a B.
– Ko te \( P(B|A) \) te tūponotanga, te tūponotanga rānei o te kaupapa B i te mea he pono a A.
– Ko te \( P(A) \) te tūponotanga o mua, te tūponotanga rānei o te huihuinga A e puta ana i mua i te whakaaro ki a B.
– Ko te \( P(B) \) te tūponotanga taha, te tūponotanga katoa rānei o te huihuinga B.

Kua Whakamaramahia nga Waahanga

1. Tūponotanga o Muri \( P(A|B) \) : Koinei te tūponotanga hou o te whakapae i muri i te whakaaro ki ngā taunakitanga hou. I roto i ngā tono mahi maha, koinei te uara e hiahia ana mātou ki te tatau.

2. Tūponotanga \( P(B|A) \) : Ka ine tēnei i te hototahitanga o ngā taunakitanga hou ki te whakapae tuatahi. He maha ngā wā ka ahu mai i ngā raraunga o mua, i ngā tauira tatauranga rānei.

3. Tūponotanga o Mua \( P(A) \) : Koinei te taumata tuatahi o te whakapono ki te whakapae i mua i te whakaarohia o tētahi taunakitanga hou. E ahu mai ana i ngā mōhiohio kua mōhiotia i mua.

4. Tūponotanga Tawhiti \( P(B) \) : He tauwehenga whakataurite tēnei e whakarite ana kia eke te tapeke o ngā tūponotanga ki te 1. Ko te tikanga ka tatauhia mā te whakaaro ki ngā huarahi katoa ka puta ai a B.

Ngā Whakamahinga Whaihua o te Ariā a Bayes

Te Tohu Hauora

Ko tētahi o ngā whakamahinga tino noa o te Ariā a Bayes ko te tātaritanga hauora. Hei tauira, me kī ka whakamātauhia tētahi tūroro mō tētahi mate. Ka puta he hua pai mai i te whakamātautau, engari he mea nui kia mārama ki te tikanga o tēnei e pā ana ki te tūponotanga kei te pāngia te tūroro e te mate.

Whakaarohia te horopaki e whai ake nei:

– Ko te tūponotanga o te tangata kia pangia e te mate (tūponotanga o mua, \( P(D) \)) he 0.01, 1% rānei.
– Ko te tūponotanga o te whakamātautau pai i runga i te mate o te tangata (tūponotanga, \( P(T|D) \)) he 0.99, 99% rānei.
– Me whakatau te tūponotanga o te whakamātautau pai mēnā he mate tō te tangata (tūponotanga taha, \( P(T) \)) i runga i ngā kēhi pai me ngā kēhi kino.

Mā te whakamahi i te Ariā a Bayes, ka taea te tatau i te tūponotanga o muri \( P(D|T) \), arā, te tūponotanga o te mate mēnā he pai te hua o te whakamātautau. Ka taea e tēnei te ārahi i ngā tākuta ki te whakatau whai whakaaro mō ngā whakamātautau me ngā maimoatanga anō.

Te Whakataunga Ture

I roto i ngā horopaki ture, ka taea te whakamahi i te Ariā a Bayes hei aromatawai i te taumaha o ngā taunakitanga. Hei tauira, mēnā ka whakaaturia ngā taunakitanga taihara ki te kōti, ka taea te ine i tōna pānga ki te tūponotanga o te hara o te tangata whakapae mā te whakamahi i te Ariā a Bayes.

Learning machine

I roto i te ako mīhini, inā koa i roto i ngā whatunga Bayesian me ngā whakarōpūtanga Bayes māori, he whānuitia te whakamahinga o te Ariā a Bayes. Ka whakamahia ēnei tauira mō ngā mahi whakarōpūtanga, ko te whāinga he whakarōpū i ngā tohu raraunga ki ngā akomanga rerekē i runga i ngā āhuatanga kua kitea.

Hei tauira, ka whakamahia e ngā tātari pāme ngā whakarōpūtanga Naive Bayes hei whakatau mēnā he pāme tētahi īmēra, i runga i te aroaro o ētahi kupu. Mā te whakahou i ngā tūponotanga ina tūtakihia ngā īmēra hou, ka whakapai ake ēnei tātari i te roanga o te wā, ka tukuna he whakarōpūtanga tika ake.

Tuhinga o mua

Ka whakamahia e ngā kaipūtea me ngā kaitātari pūtea te Ariā a Bayes hei whakahou i ngā matapae me ngā tauira ina puta mai ngā raraunga ōhanga hou. Hei tauira, ka taea e rātou te whakahou i te tūponotanga o ngā huringa reiti huamoni i runga i ngā tohu ōhanga hou, i ngā panuitanga kaupapa here rānei.

Tauira kua oti te mahi

Me titiro tātou ki tētahi tauira māmā hei tiro i te mahi a te Ariā a Bayes.

Mehemea ka hangaia e te wheketere he widget, ā, e mōhiotia ana te auau hapa. Ka whiwhi te kaiwhakahaere wheketere i tētahi widget he hapa, ā, e hiahia ana ia ki te mōhio ki te tūponotanga i ahu mai i te Mīhini A, i te Mīhini B rānei.

– E 60% o ngā taputapu ka hangaia e te Mīhini A, engari e 3% te ōrau hapa.
– E 40% o ngā taputapu ka hangaia e te Mīhini B, engari e 5% te ōrau hapa.

Me kimi e tātou te tūponotanga i ahu mai te widget hapa i te Mīhini A (\( P(A|D) \)).

Mā te whakamahi i te ariā:

\[ P(A|D) = \frac{P(D|A) \cdot P(A)}{P(D)} \]

Kei hea:
– \( P(A) = 0.6 \) (te tūponotanga o mua o te putanga o te Mīhini A).
– \( P(D|A) = 0.03 \) (te tūponotanga o tētahi hapa mai i te Mīhini A).
– Ko te tūponotanga taha o tētahi hapa ko te \( P(D) \), e tatauhia ana penei:

\[ P(D) = P(D|A) \cdot P(A) + P(D|B) \cdot P(B) \]

\[ = (0.03 \times 0.6) + (0.05 \times 0.4) \]

\[ = 0.018 + 0.02 \]

\[ = 0.038 \]

Te whakakapi i ēnei uara ki te ariā a Bayes:

\[ P(A|D) = \frac{0.03 \times 0.6}{0.038} \]

\[ = \frac{0.018}{0.038} \]

\[ \tata ki te 0.474 \]

Nō reira, e 47.4% pea te tūponotanga i ahu mai te taputapu hapa i te Mīhini A.

Opaniraa

He taputapu nui te Ariā a Bayes e āwhina ana i te whakaaroaro tūponotanga mā te whakahou i ngā whakapono ki ngā taunakitanga hou. Ko tōna whānui i roto i ngā mara maha e whakaatu ana i tōna hiranga. Ahakoa he whakahou i te tūponotanga o ngā mate hauora, he whakapai ake i ngā tauira ako mīhini, he whakamōhio i ngā whakataunga ture, he whakatikatika rānei i ngā matapae ōhanga, ka whakaratohia e te Ariā a Bayes he anga kaha mō te whakatau i ngā whakataunga whai whakaaro, e ahu mai ana i ngā raraunga. I te mea ka piki haere te rongonui o ngā raraunga, ko te mōhio ki te whakaaroaro Bayesian ka noho tonu hei pūkenga tino nui.

Waiho i te Comment