What is very interesting about this concept is the validity and reliability of them. Why they don't talk about it! it's essential when we talk about mathematical set's of any data!
Great series from IBM in general and this instructor specifically . Slightly hopeful on the Southampton bit but if you can't dream, what's the point of it all😃
Can synthetic data be as effective as real data? Wouldn’t model getting trained with synthetic data be giving false results when used against real data?
I find it difficult to stop thinking about Martin Keen, and his prediction about Southampton's future in the Premier League. It's quite remarkable that both Southampton and Leicester will be battling it out in the Championship to regain their positions in the top tier in 2025. A great example of the problems with synthetic data.
Yes, cool stuff. We use synthetic data for tracking trucks in the field. By taking existing labeled data and transforming the truck in three dimensions to get the additional data for the model.
I am trying to learn more in the information technology/AI field. I was wondering a few things. What is defined as perfectly labeled? Is there room for interpretation there? That to me, with my minimal understanding, would open an avenue of risk as it pertains to the data. Also, would it be safe to reason that synthetic data can be generated to maximize bias? Is there a safeguard to prevent this, or is it based on the ethics of the user compiling the data? Thanks for the video and replys are greatly appreciated!
I understand now what this is and why you might want it, but I'm still missing how you use it to make a real decision? Need a case study example please
Synthetic data has been very useful in my field (gene regulatory networks; maps of interactions that affect gene expression within cells). We can't manually test the interactions of tens of thousands of genes, especially across tens/hundreds of thousands of species, so we predict them using large molecular datasets. The problem is, how can you evaluate the accuracy of a prediction algorithm if you don't know what's true or false? Synthetic data is super useful, since you can generate data with known interactions that you can compare to. Algorithms can then be ranked on how close their predictions match the synthetic dataset. A great example is the GNW DREAM Network Inference Challenge, if you want to see how they use this!
I love how Synthetic Data works. It's very, very useful. I just really worry that people will start training models on Synthetic data in scenarios in which it would be dangerous to use data that is not perfectly based in reality.
Takeaway: Made up data can be used to deal with biased real word data and can be obtained from data sources or transforming existing data by adding noise or using GANs.
Interesting, if rather simplistic. Having spent the past 5/6 years developing a synthetic police-data model, it is not easy or cheap (if time is factored in). Rows and rows of financial transactions might be easy to generate, less so, complex family groups, locations, incidents and crimes, vehicles, organisations, where these are interlinked, related and reflect real-world scenarios. Whilst IBM has some excellent tools such as i2 and Watson, the real data in those systems would be unlikely to be made available for sythesising.
The following is the reply from Copilot: Can you generate a list of EPL winners from 2024 to 2033 Copilot Using the web Sent by Copilot: I can’t predict the future, so I can’t provide a list of EPL winners from 2024 to 2033. However, I can share the most recent winners up to 2024: 2023-24: Manchester City1 2022-23: Manchester City 2021-22: Manchester City 2020-21: Manchester City 2019-20: Liverpool Manchester City has been quite dominant in recent years. Do you have a favorite team you’re hoping will break their streak?
How is this not basing later models on copies of copies of potentially incorrect data? Won't we end up with piles of structurally sound, true seeming noise eventually?
Imagine I use the latest gpt model to scrape the wiki page regarding a political view point and generate 10 new pages of slightly different content based on that. All 10 will contain the lacks, flaws and biases in the original. What does thus achieve? We already have enough examples of the language in use so it's not for that. If it's for quality facts, you're not generating synthetic facts, just copies of previously learned ones. Is it just a way to get around intellectual property laws by making copies of something no one owns?
Great video! Best of luck SFC😂
I am amazed how this dude can write backwards so perfectly
See ibm.biz/write-backwards
@@IBMTechnology aha! I knew some sorcery was involved
Have you noticed they all seem to be left handed 🧐
I like this friendly instructor
Amazing series and very classical and engrossing style of explanation... keep up the good work
What is very interesting about this concept is the validity and reliability of them. Why they don't talk about it! it's essential when we talk about mathematical set's of any data!
Really love these IBM mini lectures, they are very insightful. Helped me during my college days, and are also helpful for learning as a hobby. Thanks!
Great series from IBM in general and this instructor specifically . Slightly hopeful on the Southampton bit but if you can't dream, what's the point of it all😃
I appreciate your generous use of "slightly hopeful" 🙂
Always had a sweet spot for the saints… hope my club could give a new home for JWP
Can synthetic data be as effective as real data? Wouldn’t model getting trained with synthetic data be giving false results when used against real data?
I find it difficult to stop thinking about Martin Keen, and his prediction about Southampton's future in the Premier League. It's quite remarkable that both Southampton and Leicester will be battling it out in the Championship to regain their positions in the top tier in 2025. A great example of the problems with synthetic data.
Yes, cool stuff. We use synthetic data for tracking trucks in the field. By taking existing labeled data and transforming the truck in three dimensions to get the additional data for the model.
Sounds like you used existing real data about the trucks. How is that synthetic data? I fear I'm misunderstanding this.
You are a very good teacher. Do you have a full course on this?
Excellent!!!
I am trying to learn more in the information technology/AI field. I was wondering a few things. What is defined as perfectly labeled? Is there room for interpretation there? That to me, with my minimal understanding, would open an avenue of risk as it pertains to the data. Also, would it be safe to reason that synthetic data can be generated to maximize bias? Is there a safeguard to prevent this, or is it based on the ethics of the user compiling the data? Thanks for the video and replys are greatly appreciated!
Loved this so much wow
I think this video might have jinxed Southampton. Instead of winning the Premier league they are now getting relegated.😢
so cool, thanks
I understand now what this is and why you might want it, but I'm still missing how you use it to make a real decision? Need a case study example please
Very simply explained ...👍
What kind of transparent white board is he using to write on? Very cool. Have not quite seen this before.
Searching for what synthetic data is after the release of Deepseek AI model.
Synthetic data has been very useful in my field (gene regulatory networks; maps of interactions that affect gene expression within cells). We can't manually test the interactions of tens of thousands of genes, especially across tens/hundreds of thousands of species, so we predict them using large molecular datasets.
The problem is, how can you evaluate the accuracy of a prediction algorithm if you don't know what's true or false? Synthetic data is super useful, since you can generate data with known interactions that you can compare to. Algorithms can then be ranked on how close their predictions match the synthetic dataset. A great example is the GNW DREAM Network Inference Challenge, if you want to see how they use this!
I love how Synthetic Data works. It's very, very useful. I just really worry that people will start training models on Synthetic data in scenarios in which it would be dangerous to use data that is not perfectly based in reality.
Takeaway:
Made up data can be used to deal with biased real word data and can be obtained from data sources or transforming existing data by adding noise or using GANs.
INTERESTING.😀
Can we add regional human corruption to make synthetic data more reliable one also and should it be under noise?
Interesting, if rather simplistic. Having spent the past 5/6 years developing a synthetic police-data model, it is not easy or cheap (if time is factored in). Rows and rows of financial transactions might be easy to generate, less so, complex family groups, locations, incidents and crimes, vehicles, organisations, where these are interlinked, related and reflect real-world scenarios. Whilst IBM has some excellent tools such as i2 and Watson, the real data in those systems would be unlikely to be made available for sythesising.
nice, now I can generate data for my HIV viral load detector model at no cost
Very interesting..
Thank you so much
using the prem was the perfect hook icl
Bro did Southampton dirty😂😂
Which bord you use.. ?
The one that uses spell checker.
the dataset that has SFC as potential winner of the PL, is the first I would throw away🤣🤣🤣🤣🤣🤣
Excellent presentation and excellent instructor.
The following is the reply from Copilot:
Can you generate a list of EPL winners from 2024 to 2033
Copilot
Using
the web
Sent by Copilot:
I can’t predict the future, so I can’t provide a list of EPL winners from 2024 to 2033.
However, I can share the most recent winners up to 2024:
2023-24: Manchester City1
2022-23: Manchester City
2021-22: Manchester City
2020-21: Manchester City
2019-20: Liverpool
Manchester City has been quite dominant in recent years.
Do you have a favorite team you’re hoping will break their streak?
Programming/MetaProgramming/Hypergramming.
Hypergramming is AI created synthetic databasing.
How is this not basing later models on copies of copies of potentially incorrect data? Won't we end up with piles of structurally sound, true seeming noise eventually?
Imagine I use the latest gpt model to scrape the wiki page regarding a political view point and generate 10 new pages of slightly different content based on that. All 10 will contain the lacks, flaws and biases in the original. What does thus achieve? We already have enough examples of the language in use so it's not for that. If it's for quality facts, you're not generating synthetic facts, just copies of previously learned ones. Is it just a way to get around intellectual property laws by making copies of something no one owns?
Why is it not called a fake message that is not clear in the video..
is this the hallucinations from llms like chatgpt that everyone hates put to good use?
En france nous avons lancé " l'amalgame"
Its been all downhill since Lawrie McMenemy left. #sfc
Sadly true - and that was 45 years ago!
when he said, the years when my team won the prem title, i said, lies, all lies 😀😆
currently models are being trained on sh*****Ty Ai-generated videos. definition of downward spiral--
lol u kiddin southanpton next winners haahha..
hey am from the future 2024, and SFC is not winning the premier league, sorry😂!! educative lesson💯
Who supports Southampton? 😂😂
Hahahaha