No policy can simultaneously be fully autonomous, optimally calibrated, and helpful, proving a trilemma.
problem Proving impossibility of a policy achieving maximum helpfulness, optimal calibration, and full autonomy.
method Geometric proof showing that adding any non-affine autonomy incentive to a strictly proper scoring rule destroys strict properness.
result The Behavioral Credibility Trilemma: no policy can achieve all three goals simultaneously.
This study evaluates Algorand and Ethereum 2.0 for blockchain trilemma challenges.
problem Balancing decentralization, security, and scalability in blockchain systems.
method Comparative analysis of Algorand and Ethereum 2.0 using metrics for decentralization, scalability, and security.
result Each system has strengths in addressing the blockchain trilemma, providing insights for other blockchain technologies.
New findings show privacy affects generalization error in a non-monotonic way.
problem Privacy and robustness in distributed learning.
method Theoretical analysis and matching lower/upper bounds on algorithmic stability.
result Generalization error is non-monotonically affected by privacy, depending on noise level.
New framework formalizes RLHF trilemma: improving safety, fairness, and robustness is computationally infeasible.
problem Aligning large language models with diverse human values while maintaining computational feasibility and robustness.
method Complexity-theoretic analysis integrating statistical learning theory and robust optimization.
result Achieving both representativeness (epsilon <= 0.01) and robustness (delta <= 0.001) for global-scale populations requires super-polynomial operations.
The paper proves ADL mechanisms face a trilemma and optimizes them for fairness, revenue, and exchange solvency.
problem The impossibility of a perpetual futures exchange achieving solvency, revenue, and fairness.
method Formal model of ADL, proving trilemma, and analyzing three ADL mechanisms.
result Optimized ADL mechanisms can reduce trader losses while maintaining exchange solvency.
Energy policy in Europe has been driven by the three goals of security of supply, economic competitiveness and environmental sustainability, referred to as the energy trilemma. Although there are clear conflicts within the trilemma, member countries have acted to facilitate a fully integrated European electricity marke…
Paper introduces exact credible sets for classification problems.
problem No general way to construct exact credible sets for classification.
method Generalized credible set with connection to Neyman--Pearson lemma and randomized decision rule.
result Achieves any preassigned credible level for classification problems.
Transformer architecture improved with credibility mechanism for better model performance.
problem Improving predictive models in tabular data.
method Introducing a credibility mechanism to the Transformer architecture.
result Credibility Transformer leads to superior predictive models compared to state-of-the-art models.
Peer effects, in which the behavior of an individual is affected by the behavior of their peers, are posited by multiple theories in the social sciences. Other processes can also produce behaviors that are correlated in networks and groups, thereby generating debate about the credibility of observational (i.e. nonexper…
Private statistics estimation faces a bias, accuracy, and privacy trilemma.
problem Balancing privacy, accuracy, and bias in statistical estimation.
method Use differential privacy (DP) for private statistics, but clip samples to control sensitivity and add noise for privacy, introducing bias.
result No algorithm can simultaneously have low bias, low error, and low privacy loss for arbitrary distributions.
The paper proposes a method to assess survey data credibility without needing many samples, regardless of data dimension.
problem Assessing the credibility of survey data across different dimensions.
method Task-based approach and model-specific distance metric for verifying survey data credibility in regression models.
result The sample complexity of the proposed algorithm is independent of the data dimension, making it more efficient.
Prediction markets can shape political behavior through persistent signals, not just forecast accuracy.
problem The role of prediction markets beyond forecasting.
method Transaction-level evidence from the 2024 U.S. presidential election, Signal Credibility Index (SCI).
result Price signals in prediction markets are more influential due to persistence, breadth of trader types, and cross-platform consensus.
Online reviews provide viewpoints on the strengths and shortcomings of products/services, influencing potential customers' purchasing decisions. However, the proliferation of non-credible reviews -- either fake (promoting/ demoting an item), incompetent (involving irrelevant aspects), or biased -- entails the problem o…
Credibility theory provides tools to obtain better estimates by combining individual data with sample information. We apply the Credibility theory to a Uniform distribution that is used in testing the reliability of forecasting an interest rate for long term horizons. Such empirical exercise is asked by Regulators (CRR…
BCPO optimizes offline RL policies by converting uncertainty into conservative bounds.
problem Offline RL's fragility under distribution shifts and model errors.
method Bayesian approach with credible lower bounds and KL regularization.
result BCPO yields an uncertainty-calibrated policy that avoids exploiting model errors.
CP4SBI improves the calibration of credible sets in SBI models.
problem Inaccurate credible sets in SBI models lead to underestimation of true parameters.
method Develops a local conformal calibration framework for SBI models.
result Improves the quality of uncertainty quantification for neural posterior estimators.
This project demonstrated a methodology to estimating cooperate credibility with a Natural Language Processing approach. As cooperate transparency impacts both the credibility and possible future earnings of the firm, it is an important factor to be considered by banks and investors on risk assessments of listed firms.…
In many settings, it is important that a model be capable of providing reasons for its predictions (i.e., the model must be interpretable). However, the model's reasoning may not conform with well-established knowledge. In such cases, while interpretable, the model lacks \textit{credibility}. In this work, we formally …
New auction design uses statistical learning to reduce costs and improve fairness.
problem Designing efficient multi-item auctions with reduced implementation costs and fairness.
method Nonparametric density estimation for credible intervals, two new strategies.
result Strategies consistently outperform alternative methods in revenue maximization and cost reduction.
In situations like tax declarations or analyzes of household budgets we would like to automatically evaluate credibility of exogenous variable (declared income) based on some available (endogenous) variables - we want to build a model and train it on provided data sample to predict (conditional) probability distributio…
Develops a new method for sampling from Bayesian credible sets using deep generative quantile learning.
problem Sampling from posterior distributions in high-dimensional spaces with intractable likelihoods.
method Uses deep neural networks to implicitly sample from Bayesian credible sets via a push-forward mapping and Monge-Kantorovich depth.
result Demonstrates improved performance and theoretical consistency of the quantile learning framework.
New method for credible intervals of Covid19 reproduction number.
problem Lack of credibility intervals in existing estimates.
method Combines Langevin Monte Carlo with Proximal operators.
result Produces credible intervals for reproduction number estimates.
The paper analyzes uncertainty quantification in sparse Gaussian process regression with a Brownian motion prior.
problem Analyzing uncertainty in sparse Gaussian process regression with a Brownian motion prior.
method Theoretical guarantees and limitations for pointwise credible sets are derived for a rescaled Brownian motion prior with a sparse variational Gaussian process method.
result Theoretical characterization of asymptotic frequentist coverage for credible sets, distinguishing conservative and overconfident cases.
With an eye towards human-centered automation, we contribute to the development of a systematic means to infer features of human decision-making from behavioral data. Motivated by the common use of softmax selection in models of human decision-making, we study the maximum likelihood parameter estimation problem for sof…
Recent explainability related studies have shown that state-of-the-art DNNs do not always adopt correct evidences to make decisions. It not only hampers their generalization but also makes them less likely to be trusted by end-users. In pursuit of developing more credible DNNs, in this paper we propose CREX, which enco…
The paper tests the credibility of public and private surveys using linear regression and differential privacy.
problem Ensuring the validity of data analysis results from sample surveys using linear regression.
method Designing an algorithm to test the credibility of surveys and extending it to handle LDP.
result The algorithm achieves optimal estimation error bound for ℓ1 linear regression and reduces sample complexity. Media seems to have become more partisan, often providing a biased coverage of news catering to the interest of specific groups. It is therefore essential to identify credible information content that provides an objective narrative of an event. News communities such as digg, reddit, or newstrust offer recommendations,…
Paper introduces SCI to distinguish market signals from coordination.
problem Unclear signals in prediction markets.
method Formalizes SCI, introduces weighted and time-varying extensions.
result Discriminates between market signals and coordination.
One of the major hurdles preventing the full exploitation of information from online communities is the widespread concern regarding the quality and credibility of user-contributed content. Prior works in this domain operate on a static snapshot of the community, making strong assumptions about the structure of the dat…
Online health communities are a valuable source of information for patients and physicians. However, such user-generated resources are often plagued by inaccuracies and misinformation. In this work we propose a method for automatically establishing the credibility of user-generated medical statements and the trustworth…
BIGUE algorithm provides credible intervals for hyperbolic network embeddings.
problem Uncertainty in hyperbolic network embeddings.
method Markov chain Monte Carlo (MCMC) algorithm for Bayesian hyperbolic random graph model.
result Samples from the posterior distribution provide credible intervals for hyperbolic coordinates and network properties.
New model reduces sampling cost in diffusion models, making them faster and applicable to real-world applications.
problem Challenges in generating high-quality samples, mode coverage, and fast sampling in deep generative models.
method Proposes denoising diffusion GANs that model each denoising step using a multimodal conditional GAN to reduce sampling cost.
result Demonstrates 2000imes faster sampling on CIFAR-10 dataset while maintaining competitive sample quality and diversity. Bayesian method improves estimation of unseen species.
problem Estimating unseen species in biological and physical sciences.
method Bayesian nonparametric approach with Pitman-Yor prior and Gaussian credible intervals.
result Improves asymptotic credible intervals for unseen species estimation.
Gaussian process (GP) regression is a powerful interpolation technique due to its flexibility in capturing non-linearity. In this paper, we provide a general framework for understanding the frequentist coverage of point-wise and simultaneous Bayesian credible sets in GP regression. As an intermediate result, we develop…
The paper analyzes strategic irreversible investments with novel dynamic strategies.
problem Tradeoff between preemption incentives and option value of waiting in oligopolistic markets.
method Developed novel Markov perfect equilibrium to handle singular control of optimal investment.
result Simpler strategies lead to a 'preemption trap' with zero net present values.
Underspecified ML models can behave unpredictably in real-world use.
problem ML models can fail in real-world deployment due to ambiguous predictors.
method Identified underspecification as the cause, showing it affects various ML domains.
result Underspecified models can behave differently in deployment domains.
Purpose: Malicious web domain identification is of significant importance to the security protection of Internet users. With online credibility and performance data, this paper aims to investigate the use of machine learning tech-niques for malicious web domain identification by considering the class imbalance issue (i…
There is considerable debate whether the domestic political institutions (specifically, the country s level of democracy) of the host developing country toward foreign investors are effective in establishing the credibility of commitments are still underway, researchers have also analyzed the effect of international in…
Generative sampler learns velocity fields for efficient posterior inference.
problem Sampling from complex posterior distributions in high dimensions.
method Generative multivariate posterior sampler via flow matching, learning a velocity field for a deterministic transport map.
result Conditional Brenier map enables fast generation of credible sets with theoretical consistency guarantees.
Bayesian inference engines improve density estimation accuracy and scalability.
problem Constructing accurate and scalable probability density functions.
method Bayesian inference engines (no-U-turn sampling and expectation propagation) with binning strategy.
result Density estimates have excellent comparative performance and scale well to large sample sizes.
Develops an inverse particle filter for cognitive systems.
problem Tracking cognitive adversaries in counter-adversarial applications.
method Global filtering approach using Monte Carlo methods and differentiable I-PF.
result Demonstrates convergence to optimal inverse filter and improved estimation performance.
To meet the Basel II regulatory requirements for the Advanced Measurement Approaches in operational risk, the bank's internal model should make use of the internal data, relevant external data, scenario analysis and factors reflecting the business environment and internal control systems. One of the unresolved challeng…
Novel quasi-Bayesian method for IV regression using machine learning models.
problem Uncertainty quantification in IV regression with machine learning models.
method Quasi-Bayesian procedure based on kernelized IV models and dual formulation.
result Established minimax optimal contraction rates and scalable inference algorithm.
A new portfolio model considers investor aversion to loss and risk.
problem Constructing a robust portfolio under uncertain asset returns and investor aversion.
method Distributional robust optimization (DRP) with a Wasserstein ball centered on empirical distribution, mixed-integer quadratic programming, and hybrid algorithm.
result Empirical testing shows superior performance in asset allocation compared to common strategies.
Most machine learning classifiers give predictions for new examples accurately, yet without indicating how trustworthy predictions are. In the medical domain, this hampers their integration in decision support systems, which could be useful in the clinical practice. We use a supervised learning approach that combines E…
A new method to estimate local volatility from high-frequency data.
problem Quantitative trading risk management needs a better way to estimate volatility.
method Realized local volatility surface estimated via high-frequency data and Bayesian nonparametric estimation.
result The method can capture counterfactual volatility and improve risk management.
The paper proposes a fair and private decentralized deep learning framework.
problem Ensuring fairness and privacy in collaborative deep learning.
method A reputation system and differential privacy are used. FDPDDL framework is built with two stages: initialisation and update.
result FDPDDL achieves high fairness, comparable accuracy to centralised and distributed frameworks, and better accuracy than standalone.
Bayesian method for estimating ATE with robustness to model misspecification.
problem Estimating average treatment effects under unconfoundedness.
method Double robust Bayesian inference using adjusted prior and posterior distributions.
result Bayesian credible sets form asymptotically exact confidence intervals.