text stringlengths 0 1.96k |
|---|
"But equation (2) shows a loss with no weighting." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"I'm assuming the text is correct, but then a beta should be added to the equation in front of the KL divergence." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"Tables and figures are inconveniently far from where they are referenced in the text" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']" "paper quality" |
"The weighting of the KL that the authors introduce is going to bias the learned generator towards the high probability regions." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"I.e., there are two different probabilistic models modeling the same data in inconsistent ways and one or the other is used depending on the part of the system" "['non', 'non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'a... |
"As an example, q(z) could be arbitrarily multimodal as far as the encoder is concerned, but the Weibull seems to force one mode per class." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'n... |
"But regardless of this, both models are inconsistent .)" "['non', 'non', 'non', 'non', 'non', 'arg', 'arg', 'arg', 'arg', 'non', 'non']" "paper quality" |
"Similarly, the proposed rejection sampling scheme of OCDVAE is not consistent with the theory of VAEs and it's a post-hoc tweak that is not theoretically expected to provide a pdf of data with lower KL divergence to the true data pdf" "['non', 'non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg'... |
"This paper proposes studying adversarial examples from the perspective of Bayes-optimal classifiers." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"This demonstrates that even when the Bayes-optimal classifier is robust, we may need to explicitly regularize/incentivize neural networks to learn the correct decision boundary." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non'... |
"Previously, all studies of this sort had to be done with small-scale classifiers and simplistic datasets such as Gaussians." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"The paper also definitively proves that there are realistic datasets where the Bayes-optimal classifier is non-robust, which goes against quite a bit of conventional wisdom in the field and opens up many new paths for research" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg'... |
"However, there are a few (in my opinion) critical concerns that currently bar me from strongly recommending acceptance of the paper" "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"I outline these below." "['non', 'non', 'non', 'non', 'non']" "paper quality" |
"Prior work: the paper seems to ignore a plethora of prior work around studying adversarial robustness and understanding its roots" "['non', 'non', 'non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']" "paper quality" |
"For example, a few very closely related works are as follows: - Adversarial examples are not Bugs, they are Features (pseudo-url): Ilyas et al (2019) demonstrate that adversarial perturbations are not in meaningless directions with respect to the data distribution, and in fact a classifier can be recovered from a labe... |
"While not in conflict with this work, it does closely relate and discuss many of the same issues discussed in this work, so relating them would be fruitful" "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg',... |
"Interestingly, they also construct a dataset where they Bayes-optimal classifier is robust and neural networks *do* learn a robust classifier (adversarial squares sans label noise)." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', '... |
"While I think the datasets presented in this work are much more interesting and certainly more realistic , this work should be put in context" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']... |
"Excessive Invariance causes Adversarial Vulnerability (pseudo-url): Jacobsen et al offers an explanation for adversarial examples based on the fact that NNs are not sensitive to many task-relevant changes in inputs, which seems to tie in nicely to the discussion in this paper, as under the presented setup the Bayes-op... |
"And it seems to have very relevant connections to your work." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"In fact, if real-world datasets end up being like the asymmetric dataset, then the results of this paper would actually indicate the *opposite* of the above statement" "['non', 'non', 'non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'a... |
"It is unclear on what basis one can say that real-world datasets are more like the symmetric case or the asymmetric case" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']" "paper quality" |
"I believe a more measured conclusion (perhaps that we *need* more regularization methods, but even then we may not be able to get perfect robustness and accuracy) would better fit the strong results presented in the paper" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'ar... |
"The paper justifies the adversarial vulnerability of the Linear SVM by arguing that the Bayes-optimal classifier is not in the Linear SVM hypothesis class, which makes sense" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'a... |
"The RBF SVM, for small enough bandwidth can express any function and is convex, so no argument needs to be made about its ability to find the Bayes-optimal classifier." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', '... |
"For CNNs, however, it is unclear if the Bayes-optimal classifier lies in the hypothesis class (there are ""universal approximation"" arguments but these usually require arbitrarily wide networks and are non-constructive)couldn't it be that the CNNs used here is in the same boat as the Linear SVM (i.e. the Bayes-optima... |
"Experimental setup: - One somewhat concerning (but perhaps unavoidable) thing about the experimental setup is that all the considered datasets are not perfectly linearly separable , i.e. the Bayes-optimal classifier has non-zero test error in expectation, and moreover the data variance is full-rank in the embedded spa... |
"This is in stark contrast to real datasets, where there seem to be many different ways to perfectly separate say, dogs from cats, and the variance of the data seems to be very heavily concentrated in a small subset of directions" "['non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'ar... |
"I am concerned that these properties are what drive the Bayes-optimal classifier for the symmetric dataset to be robust (concretely, if 0.01 * Identity was not added to the covariance matrix of the symmetric model and the covariance was left to be low-rank, then any classifier which was Bayes-optimal along the positiv... |
"It is unclear if what is lacking from the NN is explicit regularization, or just more data." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"While completely alleviating this concern may once again be quite difficult/impossible , it could be significantly alleviated by generating training samples dynamically (at every iteration) instead of generating a dataset in one shot and training on it" "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', ... |
"It would be very interesting to see whether these results differ at all from the one-shot approach here." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"A suggestion rather than a concern and not impacting my current score: but it would be very interesting to see what happens for robustly trained classifiers on the symmetric and asymmetric datasets." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non'... |
"I would be more than happy to significantly improve my score if these concerns can be addressed in the revision and corresponding rebuttal." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper ... |
"The dataset is an extension of CLEVR using simple motions of primitive 3D objects to produce videos of primitive actions (e.g. pick and place a cube), compositional actions (e.g. ""cone is rotated during the sliding of the sphere""), and finally a 3D object localization tasks (i.e. where is the ""snitch"" object at th... |
"A variety of models from recent work are evaluated on the three proposed tasks, demonstrating the validity of the above motivation for the construction of the dataset." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', '... |
"The primitive action classification task is ""solved"" by nearly all methods and only serves for debugging purposes." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"Finally, the localization task is challenging, especially when camera motion is introduced, with much space for improvement left for future work." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'n... |
"I am positive with respect to acceptance of this paper" "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"The authors recognize that since the dataset is synthetically generated it is not necessarily predictive of how methods would perform with real-world data, but still it can serve a useful and complementary role similar to the one CLEVR has served in image understanding" "['non', 'non', 'non', 'non', 'non', 'non', 'non... |
"I have a few minor comments / questions / editing notes that would be good to address: - The random baseline isn't described in the main text , it would be good to briefly mention it (this will also help to clarify why the value is particularly high for tasks 1 and 2) - The grid resolution ablation results presented i... |
"As this direction (of increased resolution to make the problem less artificial) is likely to be important, a brief discussion of this finding from the main paper text would be appropriate - p3 resiliance -> resilience - p4 objects is moved -> object is moved - p6 actions itself -> actions themselves; builds upon -> bu... |
"This paper proposes A*MCTS, which combines A* and MCTS with policy and value networks to prioritize the next state to be explored." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper qu... |
"It further establishes the sample complexity to determine optimal actions." "['non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non', 'non']" "paper quality" |
"Experimental results validate the theoretical analysis and demonstrate the effectiveness of A*MCTS over benchmark MCTS algorithms with value and policy networks" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']" "paper ... |
"Experimental results show that the proposed algorithm outperform the MCTS algorithms" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']" "paper quality" |
"Cons: However, there are several issues that should be addressed including the presentation of the paper : The algorithm seeks to combine A* search with MCTS (combined with policy and value networks), and is shown to outperform the baseline MCTS method." "['non', 'non', 'non', 'non', 'arg', 'arg', 'arg', 'arg', 'arg',... |
"However, it does not clearly explain the key insights of why it could perform better" "['non', 'non', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']" "paper quality" |
"How could it improve over the traditional tree policy (e.g., UCT) for the selection step in MCTS" "['arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg', 'arg']" "paper quality" |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.