I 100% understand, but the point is that some one who wrote the book on SMOTE, who worked for US Army High Performance Unit and Raytheon, who has numerous research papers about slowly adjusting data sets, was linked to a burner LinkedIn account attached to Dominion.
Forget SMOTE as the exact tech... would this some with this level have the chops to pull this off? I don't think there is a more ideal fit here.
So I do understand what SMOTE is, I've watched all his talks and read his papers before I published anything... do you think someone like this should be involved in vote tabulating software?
I'm not familiar with his whole body of work, I'm just basing my opinion on this one paper, so if you have other papers of his that might be more relevant, please post them, I'd be interested in analyzing them.
But based on this one paper, I'd say he's working on pretty mundane stuff as far as machine learning is concerned. It's basically statistically sampling techniques for training classifiers. It's foundational machine learning stuff that probably thousands of researchers are working on every day. it doesn't tell us a whole lot about his expertise in voting or anything like that.
To be more concrete to others reading this post who may be less technically inclined, the kind of manipulation the paper is talking about is limited to training data preparation for machine learning algorithms. It's no more complicated than writing an Excel macro to change some data in a spreadsheet. It's the kind of mundane data manipulation that anyone with a computer can do, so it doesn't tell us anything about whether they are an expert at manipulating voting systems.
To be fair, the math behind re-weighting vote counts is pretty mundane too. If your first batch is at 40% Biden and 70% Trump, but you want to end up at 51% Biden and 49% Trump, that’s just a simple algebra problem.
The hard part is doing it secretly and with enough randomness that it seems legitimate.
That would be easy to spot the fraud then... based on his research, the machine learning algo uses the data to balance the votes in a way that can't be detected.
I 100% understand, but the point is that some one who wrote the book on SMOTE, who worked for US Army High Performance Unit and Raytheon, who has numerous research papers about slowly adjusting data sets, was linked to a burner LinkedIn account attached to Dominion.
Forget SMOTE as the exact tech... would this some with this level have the chops to pull this off? I don't think there is a more ideal fit here.
So I do understand what SMOTE is, I've watched all his talks and read his papers before I published anything... do you think someone like this should be involved in vote tabulating software?
I'm not familiar with his whole body of work, I'm just basing my opinion on this one paper, so if you have other papers of his that might be more relevant, please post them, I'd be interested in analyzing them.
But based on this one paper, I'd say he's working on pretty mundane stuff as far as machine learning is concerned. It's basically statistically sampling techniques for training classifiers. It's foundational machine learning stuff that probably thousands of researchers are working on every day. it doesn't tell us a whole lot about his expertise in voting or anything like that.
To be more concrete to others reading this post who may be less technically inclined, the kind of manipulation the paper is talking about is limited to training data preparation for machine learning algorithms. It's no more complicated than writing an Excel macro to change some data in a spreadsheet. It's the kind of mundane data manipulation that anyone with a computer can do, so it doesn't tell us anything about whether they are an expert at manipulating voting systems.
To be fair, the math behind re-weighting vote counts is pretty mundane too. If your first batch is at 40% Biden and 70% Trump, but you want to end up at 51% Biden and 49% Trump, that’s just a simple algebra problem.
The hard part is doing it secretly and with enough randomness that it seems legitimate.
That would be easy to spot the fraud then... based on his research, the machine learning algo uses the data to balance the votes in a way that can't be detected.