M.Sc Research

[Esh95]

M.Sc in Computer Science Research කරගන්න place එකක් දන්නවද මචන්? මගේ research එක Deep learning subject related.

 

[Solo Rider]

UoC, UoM, UoK, SLIIT

Ona tharan thiyenne

 

[kasun_kbp]

සල්ලි වලටද නැත්නම් Genuinely වැඩේ බැහැලම කරන්නද?
මොනවා වුනත් උබ වැරද්දක් කරනවා මචං.
උබ ඕක කරාම ලංකාවේ එවුන් සල්ලි හොයන්නේ කොහොමද?
UoM තමා කරන්න වෙන්නේ උඩ කියලා තියෙනවා වගේ.

 

[Esh95]

UoC, UoM, UoK, SLIIT

Ona tharan thiyenne

සල්ලි වලට මචන්. Job ekath ekka time na. ekai. mn ATL knk widihta company ekka wada. mge msc eke last part ekata.

---

සල්ලි වලටද නැත්නම් Genuinely වැඩේ බැහැලම කරන්නද?
මොනවා වුනත් උබ වැරද්දක් කරනවා මචං.
උබ ඕක කරාම ලංකාවේ එවුන් සල්ලි හොයන්නේ කොහොමද?
UoM තමා කරන්න වෙන්නේ උඩ කියලා තියෙනවා වගේ.

සල්ලි වලට මචන්. Job ekath ekka time na. ekai. mn ATL knk widihta company ekka wada. mge msc eke last part ekata.
------ Post added on Aug 28, 2022 at 11:57 AM

 

[Solo Rider]

එහෙනන් මොහොක හරි රෙජිස්ට වෙලා, තෙසිස් එක පිටින් කරන්න.

මන් නම්බර් එකක් දෙන්නන්.
උන් කරනවා තෙසිස් සල්ලි වලට.

සෙට් වෙනවද කියල කතා කරල බලන්න

 

[Esh95]

එහෙනන් මොහොක හරි රෙජිස්ට වෙලා, තෙසිස් එක පිටින් කරන්න.

මන් නම්බර් එකක් දෙන්නන්.
උන් කරනවා තෙසිස් සල්ලි වලට.

සෙට් වෙනවද කියල කතා කරල බලන්න

Denna machn

 

[flag au:]

PM dapan

 

[kevin_levin]

M.Sc in Computer Science Research කරගන්න place එකක් දන්නවද මචන්? මගේ Topic එක Deep learning approach to detection of Money laundering.

සල්ලි වලට මචන්. Job ekath ekka time na. ekai. mn ATL knk widihta company ekka wada. mge msc eke last part ekata.

 

[Walter Hartwell White]

ඔය රිසර්ච් ටොපික් එක ඩිලීට් කරන්න, නැත්නම් පස්සේ අවුළක් එන්න බැරි නැහැ.

 

[Sandy_Fdo]

UoC, UoM, UoK, SLIIT

Ona tharan thiyenne

 

[Esh95]

ඔය රිසර්ච් ටොපික් එක ඩිලීට් කරන්න, නැත්නම් පස්සේ අවුළක් එන්න බැරි නැහැ.

e mko bn?

 

[NecroMonger]

Interesting topic. I assume you want to try to detect patterns in the placement part of the ML process? As layering will be difficult as yo will need to collate data from multiple banks/FIs ?
Maybe you can classify customers and find a cluster which matches the profile you are looking after.
Definitely an interesting topic.

 

[Esh95]

Interesting topic. I assume you want to try to detect patterns in the placement part of the ML process? As layering will be difficult as yo will need to collate data from multiple banks/FIs ?
Maybe you can classify customers and find a cluster which matches the profile you are looking after.
Definitely an interesting topic.

Thanks. Are u have any experience doing research?

 

[NecroMonger]

Thanks. Are u have any experience doing research?

I do but I don't undertake paid research. I have done similar things for my MSc but I can give you some pointers. For a start, using K-mean clusters for fraudulent transaction detection is a well studied area. You need to refine the said methodology and specialize on the AML side. Maybe include things like Social media aspect to enrich/enhance the clustering,
The biggest problem for you would be to find a suitable data set.
Be mindful, the first rule of AI/ML : Garbage in- Garbage out.

 

[Esh95]

I do but I don't undertake paid research. I have done similar things for my MSc but I can give you some pointers. For a start, using K-mean clusters for fraudulent transaction detection is a well studied area. You need to refine the said methodology and specialize on the AML side. Maybe include things like Social media aspect to enrich/enhance the clustering,
The biggest problem for you would be to find a suitable data set.
Be mindful, the first rule of AI/ML : Garbage in- Garbage out.

Thank you very much

 

[NecroMonger]

Thank you very much

So your whole project could be broken into the following steps

1. Securing a good data set ( data that has been anonymized but without affecting the quality)
2. Defining the transactions/patterns of interest ( Credit vs Debit Txn ratio, cash deposits, cashiers checks)
3. Enriching the dataset ( This is possible if the data is not anonymized as you need real data to profile against social media)
4. Define your input data set ( This will be the most difficult as we always tend to define our inputs to overfit)
5. Train your DNN.
6. Look at the clustering and repeat from 4 till you are happy with the results against the test data set.

 

[Esh95]

So your whole project could be broken into the following steps

1. Securing a good data set ( data that has been anonymized but without affecting the quality)
2. Defining the transactions/patterns of interest ( Credit vs Debit Txn ratio, cash deposits, cashiers checks)
3. Enriching the dataset ( This is possible if the data is not anonymized as you need real data to profile against social media)
4. Define your input data set ( This will be the most difficult as we always tend to define our inputs to overfit)
5. Train your DNN.
6. Look at the clustering and repeat from 4 till you are happy with the results against the test data set.

Thanks

 

[kinkon]

බම්ප්

 

[DJLEOx78]

bump. i might need a one too for a different domain. no time with work stress