3 ms·

This entire post and the debate surrounding it, is frankly stupid. What does 95% accuracy even means?? Consider face recognition, even when there is good gold s

This entire post and the debate surrounding it, is frankly stupid. What does 95% accuracy even means?? Consider face recognition, even when there is good gold standard for matching faces (human judgement, since human are good at recognizing faces), determining accuracy of Face recognition algorithms is still challenging (E.g. Megaface challenge). When it comes to a piece of text written by an author its even more difficult. There are several practical problems too, such as how do you distinguish Quotes and copy-pasted paragraphs from rest of the text. This sounds like a beginner who created a dataset, with a flawed metric. And is now going around claiming 95% accuracy, using "Deep" learning. And equally clueless commenters are hyping it up. Why stop at claiming 95%, hell even I can create a "dataset" and a "deep learning" algorithm and get 99.9%. I am not discounting that there are legitimate stylometric analysis methods, which have been peer reviewed. But please lets not hype "Deep learning for doxxing". This just sullies the real progress being made in deep learning.

firebones 10y ago
Agree. Rather than release the code, release the testing data set and/or test methodology used to make a claim of 95% accuracy; no one is harmed and it should give a good sense of the claim. I wouldn't be shocked to find that the score was achieved by overtraining on a very small and narrow data set (small number of identities), and that the model isn't generalizable. That said, identification may be a more tractable problem if you have a limited population, additional metadata for features, and normalized writing samples (comparing anonymous reviews to identified reviews, or within community posts, as opposed to trying to compare a set of anonymous tweets to an identifiable dissertation).
tzs 10y ago
> This entire post and the debate surrounding it, is frankly stupid. What does 95% accuracy even means?? Generally in supervised machine learning a claim of X% accuracy means that when tested on a large dataset for which the correct result is known and that was not part of the training dataset or validation dataset, it classified X% of that dataset correctly. Typically you gather a big dataset of labeled data and then split it randomly into training, validation, and test sets. A 50/25/25 split is common. If the learning approach you are using does not need a validation set, then 70/30 training/test is common. How reliable such an accuracy estimate is depends on how well your dataset matches the characteristics of the datasets people will be using your trained system on. His 95% accuracy report is probably reasonably reliable when his software is used on anonymous posts on the forums where he gathered his datasets. It would probably be less reliable looking at anonymous posts on, say, a bagpipe maker's forum.
- tiredofnoobs 10y ago
  Huh... of course I know definition of accuracy, and how its calculated. However even in supervised learning, accuracy is only used in very limited cases such as multi class classification. For a whole bunch of problems including the one being discussed its a poor and in some cases a biased metric. E.g. consider a heavily unbalanced problem 99% positive labels. By predicting all instances with majority label its possible to get 99% accuracy. There are several better metrics, False Accept rates, Precision Recall curves etc. Without knowing how the dataset was collected, did the "username" leaked into the dataset, etc. its impossible to evaluate such outlandish claims. The whole moral and ethical debate is non-sequitur, and harms legitimate deep learning research.
  - rspeer 10y ago
    I understand this concern. It's like the old 20 Newsgroups data set for testing classification, where supposedly you're distinguishing the topics of conversation between comp.graphics, sci.electronics, talk.politics.misc, and so on... ...but what the most effective classifiers do is memorize the names and signature blocks of people who posted in each newsgroup.

firebones 10y ago

Agree. Rather than release the code, release the testing data set and/or test methodology used to make a claim of 95% accuracy; no one is harmed and it should give a good sense of the claim. I wouldn't be shocked to find that the score was achieved by overtraining on a very small and narrow data set (small number of identities), and that the model isn't generalizable. That said, identification may be a more tractable problem if you have a limited population, additional metadata for features, and normalized writing samples (comparing anonymous reviews to identified reviews, or within community posts, as opposed to trying to compare a set of anonymous tweets to an identifiable dissertation).

tzs 10y ago

> This entire post and the debate surrounding it, is frankly stupid. What does 95% accuracy even means?? Generally in supervised machine learning a claim of X% accuracy means that when tested on a large dataset for which the correct result is known and that was not part of the training dataset or validation dataset, it classified X% of that dataset correctly. Typically you gather a big dataset of labeled data and then split it randomly into training, validation, and test sets. A 50/25/25 split is common. If the learning approach you are using does not need a validation set, then 70/30 training/test is common. How reliable such an accuracy estimate is depends on how well your dataset matches the characteristics of the datasets people will be using your trained system on. His 95% accuracy report is probably reasonably reliable when his software is used on anonymous posts on the forums where he gathered his datasets. It would probably be less reliable looking at anonymous posts on, say, a bagpipe maker's forum.

tiredofnoobs 10y ago
Huh... of course I know definition of accuracy, and how its calculated. However even in supervised learning, accuracy is only used in very limited cases such as multi class classification. For a whole bunch of problems including the one being discussed its a poor and in some cases a biased metric. E.g. consider a heavily unbalanced problem 99% positive labels. By predicting all instances with majority label its possible to get 99% accuracy. There are several better metrics, False Accept rates, Precision Recall curves etc. Without knowing how the dataset was collected, did the "username" leaked into the dataset, etc. its impossible to evaluate such outlandish claims. The whole moral and ethical debate is non-sequitur, and harms legitimate deep learning research.
- rspeer 10y ago
  I understand this concern. It's like the old 20 Newsgroups data set for testing classification, where supposedly you're distinguishing the topics of conversation between comp.graphics, sci.electronics, talk.politics.misc, and so on... ...but what the most effective classifiers do is memorize the names and signature blocks of people who posted in each newsgroup.