arXiv research

Blackwell's theorems influence modern AI through information compression and decision making.

We present a method for constructing the log-optimal portfolio using the well-calibrated forecasts of market values. Dawid's notion of calibration and the Blackwell approachability theorem are used for computing well-calibrated forecasts. We select a portfolio using this "artificial" probability distribution of market …

Differential privacy is a statistical concept that can be explained through hypothesis testing.

Four geometries govern sequential and distribution-free inference.

N-discount optimality was introduced as a hierarchical form of policy- and value-function optimality, with Blackwell optimality lying at the top level of the hierarchy Veinott (1969); Blackwell (1962). We formalize notions of myopic discount factors, value functions and policies in terms of Blackwell optimality in MDPs…

New approach for adaptive conformal inference using Blackwell's theory.

New method compares DP mechanisms beyond $(\varepsilon, δ)$ pairs.

The paper introduces Robust Correlated Equilibrium for games with time-varying costs and proposes an algorithm to achieve it.

Study extends binary omniprediction to multiclass setting with improved sample complexity.

Transforms offline greedy algorithms to online algorithms for combinatorial problems.

Paper improves Gumbel-Softmax estimator variance reduction.

We provide yet another proof of the existence of calibrated forecasters; it has two merits. First, it is valid for an arbitrary finite number of outcomes. Second, it is short and simple and it follows from a direct application of Blackwell's approachability theorem to carefully chosen vector-valued payoff function and …

New algorithms minimize regret with global costs in online learning.

A game-theoretic approach to multi-criteria ranking from ordinal data.

Derives new equations for stochastic volatility models.

Paper explores rate-preserving reductions between Blackwell approachability and no-regret learning.

Unified approach to fair online learning with stochastic contexts.

Calibrated strategies can be obtained by performing strategies that have no internal regret in some auxiliary game. Such strategies can be constructed explicitly with the use of Blackwell's approachability theorem, in an other auxiliary game. We establish the converse: a strategy that approaches a convex $B$-set can be…

We wish to compute the gradient of an expectation over a finite or countably infinite sample space having $K \leq \infty$ categories. When $K$ is indeed infinite, or finite but very large, the relevant summation is intractable. Accordingly, various stochastic gradient estimators have been proposed. In this paper, we de…

Proposes a method to stabilize Black Box Variational Inference using the James-Stein estimator.

Policy optimization on high-dimensional continuous control tasks exhibits its difficulty caused by the large variance of the policy gradient estimators. We present the action subspace dependent gradient (ASDG) estimator which incorporates the Rao-Blackwell theorem (RB) and Control Variates (CV) into a unified framework…

Optimizes predictions by recalibrating online forecasts with minimal error.

DSM on manifolds removes singularities and computes small-noise expansions.

A new variational method for SSMs improves inference efficiency.

New sampler reduces MCMC complexity for Bayesian variable selection.

Paper outlines a new mathematical language for experiments.

New estimator reduces variance in discrete random variables.

Improves survey sampling with unbiased machine learning methods.

Fibonacci Ensembles use Fibonacci weights to improve ensemble learning, inspired by natural growth patterns.

Partition functions of probability distributions are important quantities for model evaluation and comparisons. We present a new method to compute partition functions of complex and multimodal distributions. Such distributions are often sampled using simulated tempering, which augments the target space with an auxiliar…

New algorithm improves on static methods in Active Simple Hypothesis Testing.

We consider Blackwell approachability, a very powerful and geometric tool in game theory, used for example to design strategies of the uninformed player in repeated games with incomplete information. We extend this theory to "generalized quitting games" , a class of repeated stochastic games in which each player may ha…

New gradient estimators for discrete variables improve model training.

New methods for CI testing under model misspecification.

We introduce a dynamic mechanism for the solution of analytically-tractable substructure in probabilistic programs, using conjugate priors and affine transformations to reduce variance in Monte Carlo estimators. For inference with Sequential Monte Carlo, this automatically yields improvements such as locally-optimal pr…

Derives new equations for volatility models and option pricing.

New algorithm achieves both static and dynamic regret optimally against an oblivious adversary for deterministic losses.

New truthful calibration errors improve model ranking in multiclass prediction.

New RL method improves on standard discounted RL for operations research.

Formulae for 1-3 handle attachments in 4-manifolds.

We introduce local expectation gradients which is a general purpose stochastic variational inference algorithm for constructing stochastic gradients through sampling from the variational distribution. This algorithm divides the problem of estimating the stochastic gradients over multiple variational parameters into sma…

New framework ensures valid uncertainty estimates for any data stream changes.

A new privacy accountant for Gaussian differential privacy measures individual privacy losses.

We study online learnability of a wide class of problems, extending the results of (Rakhlin, Sridharan, Tewari, 2010) to general notions of performance measure well beyond external regret. Our framework simultaneously captures such well-known notions as internal and general Phi-regret, learning with non-additive global…

To address the challenge of backpropagating the gradient through categorical variables, we propose the augment-REINFORCE-swap-merge (ARSM) gradient estimator that is unbiased and has low variance. ARSM first uses variable augmentation, REINFORCE, and Rao-Blackwellization to re-express the gradient as an expectation und…

New method quantifies redundant information using information bottleneck.

We consider the "partial information decomposition" (PID) problem, which aims to decompose the information that a set of source random variables provide about a target random variable into separate redundant, synergistic, union, and unique components. In the first part of this paper, we propose a general framework for …

We present a novel method in the family of particle MCMC methods that we refer to as particle Gibbs with ancestor sampling (PG-AS). Similarly to the existing PG with backward simulation (PG-BS) procedure, we use backward sampling to (considerably) improve the mixing of the PG kernel. Instead of using separate forward a…