Viterbi Algorithm | HMM | Solved Decoding Example

[@saranganisivarajah7436]

I searched a lot of videos related to this.But this is a gem for me.Good explanation.thank you

[@nirperel8022]

Thank you very much for this video. I believe that the right (optimal) path is HHH, since it gives the probability 0.8*0.4*0.7*0.2*0.7*0.4=0.012544, while the path HCH gives the probability 0.8*0.4*0.3*0.5*0.4*0.4=0.00768. The optimal path should be read from right to left, in that way it is possible to see how did we get to the state we are in. Please correct me if I am wrong. Cheers.

[@nihalyadalam8933]

yes, even I think the final solution is HHH.

[@chiraagb]

hey is this approach correct ?

[@varungupta7226]

Yup, v1(1) should be 0.16 and not 0.32!

[@porkypig7170]

I also arrive at HHH

[@shivendunsahi]

At V3 calculation of C--> H will be P (H | C ) * P ( 3 | H) and from transition and Emission Matrix we get P ( H | C ) = 0.4 and P ( 3 | Hot ) = 0.04 . Hence P ( C --> H ) = 0.4 * 0.4 = 0.16.

[@binodsuman]

Thank you very much. You are right instead of 0.030, it should be 0.16. As we are taking max of both, hence final answer is correct. 😀

[@aakashyadav5765]

right.

[@meshackamimo1945]

Much appreciations from me--you are a gifted tutor. kindly prepare a short talk on Baum-Welch as well--or direct me to your existing videos on the same. Blessings,Prof.

[@78da48]

I could be wrong, but I think the last part getting the state after calculating condition probability might has some problem. I fee this need to be done backward becasue its conditional probability. so v3 we choose the max, H, and we got the H from H from v2 even H has lower probability than C. So v2 we choose H, and we got H at v2 from H in v1. so we choose H at v1. Thus the answer is H,H,H.

[@prasannap.j5179]

Yes you are right !

[@30sunique78]

And emmision Matrix also have little typo

[@deepikasingh3122]

Can you explain how? If u're right

[@johnmanipadam1926]

Great explanation, thanks, however, i feel that the path with the maximum probability is incorrect. the last (third) letter can be inferred by finding the max of both in third place, however the second letter has to be backtracked from the third using dynamic programming rather than just calculate the max of the value of two second letters, because the value of the third letter is the max of the product of the value of the second letter and the transition and emission matrix not just the value of the second letter.

[@chandrikamohan6950]

I loved the explanation and everything was clear. It would've been better if the camera was fixed in a particular position instead of moving it all over, that was highly distracting. Great video though!

[@mahaadel2833]

thank you so much, it makes sense now. my lecturer could not explain anything.

[@sudhirsethi497]

Excellent video Binod. Try to minimize movement of screen. Considering lot of information on a single screen difficult but min as much as possible.

[@binodsuman]

Thank you Sudhir for your nice words and feedback. I’ll try to improve. Keep Learning !!

[@swapneelsahu9440]

Your smile is so infectious in each and every video! It gives such good vibes! I have no words! 🙌🏼🙌🏼🌟🌟💯💯

[@habibabouhalouf163]

Thank you very much, but At V3 P( C--> H)=P(3 | H)*P(H | C) = 0.4*0.4=0.16 so V3(2)=max(0.007168, 0.00288)=0.007168. The optimal path does not change : HCH

[@Living_Vibes]

Thanks for this I understood the topic with ease

[@keshavsinghal8546]

This is incredibly thorough. Thank you so much!

[@bittu6724]

Sir, In emission Matrix you need to 0.2 0.4 0.2 instead of 0.2 0.4 0.4

[@mayalurianusha385]

It was really very clear and great explanation sir.👏

[@binodsuman]

Thank you so much for your nice word, it meant a lot for me to do more. Keep Learning !!

[@saikatbandopadhyay]

wonderful explanation. clear use of words. sound example. very good video to make this algorithm easy to understand.

[@Shikhar_]

The graphical representation was quite helpful. Thank you

[@imrankhan-nv7mf]

please please sir! make a video on 'Forward Probability Algorithm" as soon as possible. All your videos are simple and awesome. waiting eagerly sir for that.....

[@shruthihm6057]

Thank you Binod ! You have explained it very nicely.

[@nadianiknam938]

It was great. Would you please explain what is the difference between hidden markov and POMDP by example? And how to update belief in POMDP?

[@yashgoyal9878]

thanku sir , this save me on my paper day

[@Knud451]

Thanks a lot for this! This was a very good walkthrough. From this numerical example, could I use that as an example to measure against when implementing the Viterbi algorithm in python? I mean is there sufficient information in this example to cover a full implementation? I just want some specific numbers to test against.

[@bhim443]

Very Nice Explanation.. fully understood the concept.. thanks

[@supriyam2945]

Clearly explained... but the finding of the hidden nodes at the last is not correct... it is not finding the maximum of all the levels, it has to start from the right end, find the maximum in that level and then backtrack to the node that helped in making that entry and repeat the steps recursively till the first level to find the hidden nodes... In this case, it will start from H of V3 and then find the predecessor that helped in making at entry of 0.012 at V3 and backtrack... This procedure repeats...

[@OmarMH87]

that's correct

[@ourybah6227]

so is the right answer HHH? because that's what I think the Viterbi should output in this problem, I am confused

[@landgerupa2290]

Understood very well, Thanks a lot for nice explanation .

[@binodsuman]

Glad to know this Viterbi Algorithm HMM video helped you.

[@ronitpaul1014]

Sir your explanation was good. Can you provide its code in R

[@PraveenKumar-rd8bj]

Thank you sir ur explanantion is clear!!!!!

[@ImPushpendraPal]

thanks for the video. I think there is a mistake at c to h from v(2) to v(3). value should be 0.08 but you mistakenly written as 0.03

[@_justinxu]

Best explanation ever, thanks bro :D 😎

[@NidhiSingh-bw7sb]

🙏 excellent explanation. Sir ! Please include some more topics..n-grams, BLEU Score etc

[@srinivasadineshparupalli5139]

Awesome sir. I really understood clearly.

[@binodsuman]

Thank you for nice words. Good to know Viterbi Algorithm video helped you. Keep Learning !!

[@QuratRaja-q2v]

Worthy Explanation!

[@Chillos100]

Best explanation thus far!! Thnx a lot!

[@apoorvbhargava1745]

Very well explained. Thanks a lot

[@sobhanbabuch9230]

emission matrix table u made small mistake H-->3 is 0.2 rest of all is good thanking you

[@priyankkoul4581]

Thank you, Sir! Helped a lot.

[@mayiflex]

really well explained, thanks!

[@pawanchoure1289]

THANKS

[@tyow95]

Thank you! This was really good.

[@cauchysequence911]

Please make video about baum welch algorithm

[@llmanarll]

Thank you so much

[@binodsuman]

Thank you for your kind comment and encouragement! Your support means a lot to me and I'm grateful to have you as a viewer. I'll do my best to continue creating content that you find helpful and enjoyable.

[@pablo-z-dragon]

thank you so so much, so easy to understand, cheers!

[@deveshnandan323]

Thanks a lot Sir :)

[@nightingale7031]

Thanks Dude u help me a lot

[@learnwithpriyanshiandkhuwa9669]

Sir please provide the vedio on maximum entropy model in nlp

[@mahendrashinde7047]

sir please make video on BERT

[@dugongzzz]

super helpful! Thanks a ton!

[@vishalkuber3631]

best explanation.. keep it bro

[@sahanravindu4223]

Thank you sir

[@MoMo-di5vy]

Thank you so much.

[@kupomomo1712]

my professor didnt explain shit, thank you

[@binodsuman]

Glad to hear that this Viterbi Algorithm HMM Tutorial series helped you. Keep Learning and thank you for your nice words !!

[@jeonyouna1799]

thank you a lot

[@muragekibicho3646]

yesss! thank you!

[@AmmasKitchenFood]

Nice explanation

[@binodsuman]

Nice to hear that Viterbi Algorithm video somehow helped you. Keep Learning !!

[@ccuuttww]

No backtracking?

[@vennilat7786]

Thanks sir

[@sudhasenthilkumar335]

how did you get pi value

[@dbgm12]

Hi Sir where can I download your workings?

[@aryanbhardwaj7642]

hey king keep being you

[@amrujaakhtertusty7471]

Nice

[@mrudulshirodkar1332]

the videographer is so bad , otherwise good explanation thank you

[@reve2051]

just want to ask if that also trellis diagram?

[@williamkmp9998]

Binus ?

[@reve2051]

@@williamkmp9998 what is binus?

[@itskakeru]

@@williamkmp9998 hmm interesting, COMP6639

[@ramsahu8361]

one thing i am not getting, Is Vitebri and HMM same thing ?

[@binodsuman]

HI Ram, connect are same but way of solving is different. Using Viterbi you can solve problem very easy fast. Good to know that Viterbi Algorithm for HMM helped you to learn. Kindly let me know if you do have any more question on it. Keep Learning !!

[@ramsahu8361]

Thanks Sir 👍

[@tacklewithtricksbykajala.4231]

U didn't tell that which probability is ans higher or lower ?

[@binodsuman]

Higher probability will be the answer.

[@__________________________6910]

Thanks !

[@tigermotivation2456]

Allah'ına kurban. Thanks very much

[@OmarMH87]

It seems you find best state not best path

[@mdhabibulislam3896]

what is hatch here?

[@shubhankarraja2412]

Justice for Cameraman

[@mulyevishvesha1283]

And 4th day he get fever and cold 😝

[@binodsuman]

Ha ha 😊😊.

[@lakshmanpr9270]

the final answer should be Hot Hot Hot not Hot cold Hot can you please confirm

[@Knud451]

Cold has a higer probability in v2.

[@chidam333]

15:02

[@vaiebhavpatil2340]

give notes

[@vaddesai4254]

h c c

[@emikarunarathne4865]

Nice explanation

[@apoorvbhargava1745]

Very well explained. Thanks a lot

[@binodsuman]

Thank you. Glad to know Viterbi Algorithm HMM tutorial video helped you to learn. Keep Learning !!

[@mafujmolla5069]

thanks a lot