[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math


View post   

File: 3.99 MB, 400x225, smug2.gif [View same] [iqdb] [saucenao] [google]
10366114 No.10366114 [Reply] [Original]

What new and interesting things has /sci/ been researching?

>> No.10366126

>>10366114
Recently decided to apply to a research opportunity at my university (am a freshman, decided not to go anywhere else just yet). One project involves the implementation of AI in video games, and the other concerns the development of a new concept of diagonalization in linear algebra

>> No.10366179
File: 132 KB, 396x385, 1519074277934.png [View same] [iqdb] [saucenao] [google]
10366179

Personal research currently focuses on mostly AI stuff. Usually with applications to finance and economics. I'm really interested in how different machine learning methods can be used to solve problems on portfolio allocation/auctions/contracts/risk aversion etc. Also working in a lab at my school focused on NLP & cognitive computing applications. Going to work in a statistics lab this summer to broaden my perspectives a bit.

>> No.10366186

Hidden Markov Model for voice recognition.
I'm struggling with statistics, it's been a long time since I've done anything in the field.

>> No.10366217

Recently published a paper with nice figures in a low impact journal which will probably be discredited within a couple years.

>> No.10366218

>>10366114

How to get a girlfriend

>>10366217

Nice

>> No.10366221

>>10366217
What is the paper about? Can you link the paper here?

>> No.10366231

>>10366186
>Hidden Markov Model for voice recognition.
I'm pretty sure between Facebook, Google, Amazon and the three letter agencies they've got voice recognition perfected.

>> No.10366240

>>10366231
>perfected
Have you used auto-generated captions on YouTube? They still have a long way to go before they can recognize speech as efficiently as humans can.

>> No.10366246

>>10366221
>Can you link the paper here?
yeah sure. not

>> No.10366248

>>10366246
Can you describe the goddamn paper then?

>> No.10366264

>>10366248
It has pretty pictures.

>> No.10366299

>>10366186
Doing kinda the same. Which feature extractor do you think is most suitable, and why?

>> No.10366775
File: 29 KB, 665x574, akkoretarded.jpg [View same] [iqdb] [saucenao] [google]
10366775

>>10366264

>> No.10366788

>>10366217
>which will probably be discredited within a couple years
elaborate?