arXiv research

Characterizes sample complexity for outcome indistinguishability in machine learning.

The paper examines how to test if two learning algorithms produce similar outcomes.

A new metric evaluates classification algorithms at the point of indistinguishability.

New index principle shows indistinguishability of certain knot crossings.

New approach shows backdoor attacks are indistinguishable from natural data features.

Framework uses human judgment to distinguish algorithmically indistinguishable cases.

Survey on calibration in machine learning, viewing it as indistinguishability.

DAL improves active learning for neural networks with large batch sizes.

Efficiently generates models resistant to falsification.

The paper formalizes robot environments using topological concepts.

GCNs struggle to distinguish certain graph models, especially those with high degree profiles.

We investigate the possibility of statistical evaluation of the market completeness for discrete time stock market models. It is known that the market completeness is not a robust property: small random deviations of the coefficients convert a complete market model into a incomplete one. The paper shows that market inc…

Differential privacy has emerged as a gold standard in privacy-preserving data analysis. A popular variant is local differential privacy, where the data holder is the trusted curator. A major barrier, however, towards a wider adoption of this model is that it offers a poor privacy-utility tradeoff. In this work, we add…

Transformers approximate mean-field dynamics of indistinguishable particles.

A new approach for instance-optimal learning that bypasses impossibility results.

Generative Adversarial Networks are a new family of generative models, frequently used for generating photorealistic images. The theory promises for the GAN to eventually reach an equilibrium where generator produces pictures indistinguishable for the training set. In practice, however, a range of problems frequently p…

Study on replicability in reinforcement learning algorithms.

The possibility of statistical evaluation of the market completeness and incompleteness is investigated for continuous time diffusion stock market models. It is known that the market completeness is not a robust property: small random deviations of the coefficients convert a complete market model into a incomplete one.…

Graph neural networks fail to distinguish certain 3D atom configurations.

Unsupervised learning models can be indistinguishable without identifiability, leading to unreliable representations.

Graphs indistinguishable by GNNs are fully characterized.

Time dilation $\frac{1}{\sqrt{1-v^2}}$ and relative velocity $v$ are observationally indistinguishable in the special theory of relativity, a duality that carries over into the general theory under Fermi coordinates along a curve (in coordinate-independent language, in the tangent Minkowski space along the curve). For …

We show social events can be accurately predicted, but often undesirably.

Enhanced attacks quantify machine learning data leakage.

Efficient algorithms for deleting data from machine learning models without significantly affecting performance.

Two knots in three-space are S-equivalent if they are indistinguishable by Seifert matrices. We show that S-equivalence is generated by the doubled-delta move on knot diagrams. It follows as a corollary that a knot has trivial Alexander polynomial if and only if it can be undone by doubled-delta moves.

How can we explain the predictions of a black-box model? In this paper, we use influence functions -- a classic technique from robust statistics -- to trace a model's prediction through the learning algorithm and back to its training data, thereby identifying training points most responsible for a given prediction. To …

We establish tools to facilitate the computation and application of the Chekanov-Eliashberg differential graded algebra (DGA), a Legendrian-isotopy invariant of Legendrian knots in standard contact three-space. More specifically, we reformulate the DGA in terms of front projection, and introduce the characteristic alge…

Paper distinguishes causal structures under latent confounding and selection bias.

We tackle the problem of protein secondary structure prediction using a common task framework. This lead to the introduction of multiple ideas for neural architectures based on state of the art building blocks, used in this task for the first time. We take a principled machine learning approach, which provides genuine,…

Machine learning models leak information about the datasets on which they are trained. An adversary can build an algorithm to trace the individual members of a model's training dataset. As a fundamental inference attack, he aims to distinguish between data points that were part of the model's training set and any other…

Deep generative models are rapidly becoming a common tool for researchers and developers. However, as exhaustively shown for the family of discriminative models, the test-time inference of deep neural networks cannot be fully controlled and erroneous behaviors can be induced by an attacker. In the present work, we show…

We resolve the open problem of optimal sample complexity for multicalibration and deterministic predictors.

Paper bypasses backdoor detection algorithms in deep learning models.

This paper introduces a method to approximate Gaussian process regression by representing the problem as a stochastic differential equation and using variational inference to approximate solutions. The approximations are compared with full GP regression and generated paths are demonstrated to be indistinguishable from …

The relationship between statistical dependency and causality lies at the heart of all statistical approaches to causal inference. Recent results in the ChaLearn cause-effect pair challenge have shown that causal directionality can be inferred with good accuracy also in Markov indistinguishable configurations thanks to…

We inject undetectable backdoors into obfuscated neural networks and language models.

New graph neural networks can distinguish graphs better than previous models.

PathNNs improve graph neural networks by distinguishing non-isomorphic graphs.

Bayesian method for causal discovery from unknown general interventions.

Representation learning (RL) plays an important role in extracting proper representations from complex medical data for various analyzing tasks, such as patient grouping, clinical endpoint prediction and medication recommendation. Medical data can be divided into two typical categories, outpatient and inpatient, that h…

Generative models create indistinguishable adversarial objects for object detection.

New approach tackles decision-making under predictions that shape outcomes.

Different optimizer choices lead to different financial model predictions.

We study, using Mean Curvature Flow methods, 2+1 dimensional cosmologies with a positive cosmological constant and matter satisfying the dominant and the strong energy conditions. If the spatial slices are compact with non-positive Euler characteristic and are initially expanding everywhere, then we prove that the spat…

CRBM generates digital twins for MS patients, aiding in disease progression analysis.

In recent years, advances in machine learning algorithms, cheap computational resources, and the availability of big data have spurred the deep learning revolution in various application domains. In particular, supervised learning techniques in image analysis have led to superhuman performance in various tasks, such as…

Modified model prevents volatility from approaching zero.