HNHacker News
TopNewBestAskShowJobs

ersiees

389 karma · joined January 6, 2015

submissionscomments
ersiees··on TabPFN: Transformer Solves Small Tabular Classification in a Second
Yeah, we also find this interesting. Arguably, though, GP is far worse in all metrics. Thus, it is not really our closest competition.
ersiees··on TabPFN: Transformer Solves Small Tabular Classification in a Second
Hey there! I am one of the authors on this paper. If there are any questions, I am happy to answer :)
ersiees··on Why do tree-based models still outperform deep learning on tabular data?
Yes, good point! The paper I am responsible for is targeted at smaller datasets, but I will propose this to the authors of the second paper :)
ersiees··on Why do tree-based models still outperform deep learning on tabular data?
The first work, I linked, arguably gets around this issue completely. There we train a neural network to approximate Bayesian prediction directly. It accepts the dataset as input. Thus, there is no optimisation procedure per dataset, just a forward pass. So far, this seems to be pretty stable.
ersiees··on Why do tree-based models still outperform deep learning on tabular data?
Our lab works on changing this. I think it might still take some years for a full solution, but so far we are successful with NNs for small datasets with meta-learning (https://arxiv.org/abs/2207.01848) and large datasets with regularisation (https://arxiv.org/abs/2106.11189). The first is very new, but the second is also cited in this paper, but they didn’t run it as a baseline.
ersiees··on Planting Undetectable Backdoors in Machine Learning Models
From working in a company providing neural network inference as a service, I can attest you, that we did this. We did it especially since we are scared people distill on our results. If the other service makes the same weird mistake, they distilled from us.
ersiees··on PyTorch 1.10
There also are new application libraries (torch vision, audio, X) being released together with 1.10: https://pytorch.org/blog/pytorch-1.10-new-library-releases/
ersiees··on PocketPCR
Can I use this to run my own COVID-19 tests for cheap?
ersiees··on WhatsApp's new privacy policy is so bad it might be illegal
What a weird assumption.
ersiees··on A New Lens on Understanding Generalization in Deep Learning
The real comparison would have been between Cifar-10 (or Imagenet) and a downsampled version there of, not upsampled, and I actually know for sure that this harms performance.. So this ideal world and real world training are definitely not the same!
ersiees··on Sales of electric cars up by 43% in 2020
You might have missed the production of EVs that causes more CO2 than that of diesel cars.
ersiees··on The campaign against plastic is harming the planet and the public
I really like this critique of environmentalism, as an environmentalist myself. While I think it should have ended with an argument in favour of some sustainability law, like a carbon dioxide tax, it clearly remembers us that still we are so much stuck in ideologies when it comes to saving co2 and that this could potentially be, because it makes us feel better than really saving co2.
ersiees··on Low-code, no-model recommendation system with GPT3
You might like: https://transformer.huggingface.co The GPT-2 model is very similar to GPT-3 just smaller.
ersiees··on Data-Transfer-Project
I don’t understand. Why would Facebook or Twitter support that? Isn’t it only bringing their monopolies in danger? Could someone explain, please.
ersiees··on Ask HN: Best tablet for white boarding during a video conference?
I use an iPad (Not pro), which also supports Apple Pencil by now and costs only a few hundred euros. That works very well, too.
ersiees··on Why general artificial intelligence will not be realized
Interesting idea! Future Me: A computer became better at my job, but I might still claim that it is not generally intelligent because it does not have the same kind of relationships to humans I have.
ersiees··on The cost to train an AI system is improving at 50x the pace of Moore’s Law
I would really like a thorough analysis on how expensive it is to multiply large matrices, which is the most expensive part of a transformer training for example according to the profiler. Is there some Moore’s law or similar trend?
ersiees··on Data Submitted by the Italian Civil Defense on COVID19 in Italy, Updated Daily
I wonder if Zeit, people with machine learning background, could do to help given open data. Is there some open source project or so to contribute to that does something helpful e.g. predicting cases or areas at risk?
ersiees··on Quadratic Payments: A Primer
No, there is the letter b indicating the payoff multiplier. So you should not bet if in expectation you lose on the bet, which seems intuitive.
ersiees··on A mathematical analysis of Uniswap markets
I guess the difference here is that a decentralised order book on the block chain would be run by every node, which is why compute in the blockchain (gas) is quite expensive.
ersiees··on Facebook to integrate the infrastructure for WhatsApp, Instagram and Messenger
This would actually make me delete my Facebook account finally, if I can message all people on Facebook through WhatsApp.
ersiees··on Sweden's cashless society is no longer a utopia
It changes the value fluctuations. Right now the biggest problem with using crypto as your day to day money is that you cannot trust the money that you put into crypto to stay at the same value over time + transactions are slow.
ersiees··on Tell HN: Executing a 51% attack on a real live CryptoCurrency, livestream
What these attacks are about is stealing money, right? Is that your plan? Why do you want donations to steal something or if you don’t steal anything how will you change the chain?
ersiees··on How counterfeits benefit Amazon
It is a problem in Germany too. You might try searching for “Apple EarPods” and see if the results look trustworthy.
ersiees··on Ask HN: Best alternative to Gmail?
Why isn’t it actually possible to just encrypt saved emails on server? So that government does not have access. Couldn’t one use a hash of the password as key for the data for the data and not save that hash to check password but another one. This way (practically), at least if the password is not eavesdropped and saved by the mail provider, it would be much harder to give away emails.
ersiees··on Serverless Best Practices
You don’t need a reconnect for every invocation actually, only per instance. I know you don’t want to think about instances with lambda, but actually they just run on some VM and you can just open one connection per VM.
ersiees··on How well does population density predict U.S. voting outcomes?
Did not work on my iPhone 6 in Germany. Saw a lot of: “ RuntimeError: us could not be resolved“
ersiees··on Practical homomorphic encryption over integers (2017)
I think this is what zcash already does.
ersiees··on France’s government is building its own encrypted messenger service
What about using Signal? Will they do something else then using the Signal protocol? If so, this might be either pretty expensive or could go wrong.
ersiees··on The Amazon worker: paid £18,000 a year to shift 250 items an hour
(i) I think as pointed out in the last paragraph, the state is as guilty as Amazon in letting something like this happen. If Amazon would not bring workers to the legal/market limit. There would be another company to pick at here. (ii) No corporation cares about being ethical correct. Companies only purpose is to generate (more) profits and this can be even more easily (without moral burdens) achieved if (like in big corporations) there are a lot of layers of abstraction between the decision makers view and the workers view. (iii) I think if Amazon would be a nice company something like this would not happen.
← PreviousPage 2 of 3Next →