Study on self-consuming generative models with diverse human curation, focusing on convergence and stability.
problem Analyzing self-consuming generative models with heterogeneous human curation.
method Investigates the asymptotic behavior of retraining dynamics using nonlinear Perron--Frobenius theory and Banach contraction mapping.
result Improves convergence results and provides stability and non-stability analyses for the model.
The paper shows how curated synthetic data can optimize human preferences in generative models.
problem Contamination of web-scale datasets by synthetic data affects future model training.
method Theoretical study of iterated retraining of generative models with curated synthetic data.
result Data curation can be seen as an implicit preference optimization mechanism, maximizing expected reward.
Method curates cost-effective, high-quality datasets using AI models.
problem Costly manual labeling of datasets.
method Probably Approximately Correct Labels (PACL) method.
result Curates high-quality datasets with low overall labeling error.
PILAF optimizes reward models from human feedback for better policy alignment.
problem Creating accurate reward models from human feedback for policy optimization.
method Policy-Interpolated Learning for Aligned Feedback (PILAF) that explicitly aligns preference learning with maximizing underlying oracle reward.
result PILAF is optimal from both optimization and statistical perspectives, demonstrating strong performance in RLHF settings.
Method tackles uncertainty in reward models for LLMs from heterogeneous human feedback.
problem Uncertainty in reward models for LLMs from heterogeneous human feedback.
method Heterogeneous preference framework and alternating gradient descent algorithm.
result Established theoretical guarantees for estimator convergence and asymptotic distribution.
For assistive robots and virtual agents to achieve ubiquity, machines will need to anticipate the needs of their human counterparts. The field of Learning from Demonstration (LfD) has sought to enable machines to infer predictive models of human behavior for autonomous robot control. However, humans exhibit heterogenei…
LoCo-RLHF models diverse human feedback with contextual information.
problem Heterogeneous human feedback from diverse contexts and preferences.
method Low-rank contextual preference model, PRS policy.
result LoCo-RLHF achieves tighter sub-optimality gap than existing methods.
FinHEAR combines LLMs with human expertise for better financial decision-making.
problem Challenges in financial decision-making for language models.
method Multi-agent framework with specialized LLMs for historical analysis, event interpretation, and expert retrieval.
result FinHEAR outperforms baselines in financial tasks with higher accuracy and risk-adjusted returns.
We explain SSL objectives as log-likelihoods in a data curation model.
problem Lack of understanding of SSL objectives as log-likelihoods.
method Formulate SSL objectives as a log-likelihood in a generative model of data curation.
result SSL methods can be understood as lower-bounds on a principled log-likelihood.
The development of human-robot systems able to leverage the strengths of both humans and their robotic counterparts has been greatly sought after because of the foreseen, broad-ranging impact across industry and research. We believe the true potential of these systems cannot be reached unless the robot is able to act w…
The paper examines how decentralized credit curators have taken over risk management from traditional protocols.
problem Risk management in decentralized credit has shifted from centralized protocols to decentralized curators.
method Analysis of ERC 4626 vaults and third-party curators, focusing on capital utilization, concentration, and fee margins.
result Curators have a significant impact on the risk profile of decentralized credit systems, with a small set of curators handling a disproportionate share of system TVL.
The paper investigates ethical issues in large image datasets, focusing on pornographic content.
problem Ethical issues in large-scale computer vision datasets, particularly concerning pornographic content.
method Cross-sectional model-based quantitative census covering various factors in the ImageNet-ILSVRC-2012 dataset.
result The dataset contains verifiably pornographic images, including non-consensual and voyeuristic content.
The paper resolves the paradox of using less data in machine learning.
problem The paradox of using less data in machine learning.
method Theoretical framework and data curation strategies.
result Small curated datasets can outperform full datasets under certain conditions.
VHGM-MAE generates synthetic humans from healthcare data.
problem Handling high-dimensional, sparse healthcare data with missing values.
method Masked autoencoder (MAE) tailored for healthcare data, addressing heterogeneity, missingness, and high-dimensionality.
result VHGM-MAE outperforms existing methods in missing value imputation and synthetic data generation.
FLAIR dataset for federated learning benchmarks.
problem Lack of suitable federated learning datasets.
method Curated large-scale annotated image dataset for multi-label classification.
result FLAIR captures real-world federated learning challenges.
Proposes measuring fairness through multiple stakeholder-curated stress tests.
problem Limited power of rigid fairness metrics and lack of stakeholder involvement in fairness discussions.
method Shift focus from fairness metrics to stress tests curated by stakeholders.
result Machine's performance under multiple stress tests reflects fairness.
Reinforcement learning (RL) has achieved tremendous success as a general framework for learning how to make decisions. However, this success relies on the interactive hand-tuning of a reward function by RL experts. On the other hand, inverse reinforcement learning (IRL) seeks to learn a reward function from readily-obt…
We report the first, to the best of our knowledge, hand-in-hand collaboration between human rights activists and machine learners, leveraging crowd-sourcing to study online abuse against women on Twitter. On a technical front, we carefully curate an unbiased yet low-variance dataset of labeled tweets, analyze it to acc…
This work aims to reduce inexplicable errors in deep neural networks by obtaining class-level semantics and penalizing misclassifications.
problem Deep neural networks misclassify images, leading to inexplicable errors that can harm trust and societal impact.
method Obtain class-level semantics, propose Weighted Loss Functions (WLFs), and train classifiers with these methods.
result Trained networks have more explicable failure modes and comparable accuracy to existing methods.
TransNet improves community detection on target networks using privacy-preserved source networks.
problem Community detection on sensitive network data with privacy constraints.
method Spectral clustering framework leveraging locally distributed privacy-preserved auxiliary networks via randomized response and adaptive weighting.
result TransNet delivers strong gains in community detection across various privacy levels and heterogeneity patterns.
Resource scheduling and coordination is an NP-hard optimization requiring an efficient allocation of agents to a set of tasks with upper- and lower bound temporal and resource constraints. Due to the large-scale and dynamic nature of resource coordination in hospitals and factories, human domain experts manually plan a…
Recent developments in Neural Relation Extraction (NRE) have made significant strides towards Automated Knowledge Base Construction (AKBC). While much attention has been dedicated towards improvements in accuracy, there have been no attempts in the literature to our knowledge to evaluate social biases in NRE systems. W…
Autonomous systems can substantially enhance a human's efficiency and effectiveness in complex environments. Machines, however, are often unable to observe the preferences of the humans that they serve. Despite the fact that the human's and machine's objectives are aligned, asymmetric information, along with heterogene…
DataRater learns which data points are most valuable for training models.
problem Training model efficiency depends on high-quality training data.
method Meta-learning to estimate the value of data points for training.
result Meta-learning improves compute efficiency by filtering data effectively.
The paper studies how to allocate human validation in AI-assisted tasks to minimize errors.
problem Heterogeneous reliability of AI-generated signals across tasks, products, and customer segments.
method Tuned prediction-powered inference, upper confidence bounds policy, Neyman square-root rule.
result The proposed policy outperforms uniform and epsilon-greedy allocation, closing most of the gap to the oracle when reliability is heterogeneous.
In our work, we propose to represent HTM as a set of flat models, or layers, and a set of topical hierarchies, or edges. We suggest several quality measures for edges of hierarchical models, resembling those proposed for flat models. We conduct an assessment experimentation and show strong correlation between the propo…
This work simplifies data valuation for LLMs using Shapley value computation.
problem How to fairly distribute benefits from training superior LLMs with multiple data owners' resources.
method We leverage the specific mathematical structure of DPO to enable scalable Shapley value computation for LLMs.
result We demonstrate that Shapley value computation for LLMs trained with DPO is significantly simplified.
We provide an online RLHF workflow for large language models.
problem Lack of online RLHF workflow for large language models.
method Construct preference models using open-source datasets, approximate human feedback, and implement online iterative RLHF.
result Our LLM achieves state-of-the-art performance on various benchmarks.
The recent advances in Natural Language Processing have been a boon for well-represented languages in terms of available curated data and research resources. One of the challenges for low-resourced languages is clear guidelines on the collection, curation and preparation of datasets for different use-cases. In this wor…
The paper compares aggregated data labels in curated and random bags for machine learning models.
problem Protecting user privacy in machine learning systems with aggregated data.
method Examined curated and random bags for training machine learning models and compared their performance.
result Gradient-based learning can be performed on aggregated data without performance degradation.
Plud system reduces labeling time and produces accurate models for uncategorized images.
problem Time-consuming manual labeling of uncategorized images.
method Iterative semi-supervised workflow combining clustering and classification.
result Reduces labeling time and produces accurate models.
LARP filters data to protect model performance across various learners.
problem Protecting model accuracy in public datasets with diverse learners.
method Formalizes and analyzes LARP, a robust data prefiltering method.
result LARP provides guarantees on worst-case loss over a set of learners, with some performance trade-off.
Response time improves alignment with diverse human preferences.
problem Standard aggregation of feedback ignores heterogeneity and anonymity.
method Augmenting feedback with response time data and modeling decisions with DDM.
result Estimator of heterogeneous preferences converges to true average preference.
Negative screening is one method to avoid interactions with inappropriate entities. For example, financial institutions keep investment exclusion lists of inappropriate firms that have environmental, social, and government (ESG) problems. They create their investment exclusion lists by gathering information from variou…
For a social networking service to acquire and retain users, it must find ways to keep them engaged. By accurately gauging their preferences, it is able to serve them with the subset of available content that maximises revenue for the site. Without the constraints of an appropriate regulatory framework, we argue that a…
Machine learning research depends on objectively interpretable, comparable, and reproducible algorithm benchmarks. We advocate the use of curated, comprehensive suites of machine learning tasks to standardize the setup, execution, and reporting of benchmarks. We enable this through software tools that help to create an…
Training models to prefer certain responses can unintentionally shift probability to harmful ones.
problem Likelihood displacement in DPO models, leading to unintended unalignment.
method Characterized and mitigated likelihood displacement using CHES score.
result Training models to prefer certain responses can unintentionally shift probability mass to harmful responses.
Cold posteriors in BNNs harm performance, likely due to incorrect likelihood.
problem Cold posteriors in Bayesian neural networks degrade performance.
method Developed a generative model explaining cold posteriors and matched it to the tempered likelihoods.
result Cold posteriors are a result of using the wrong likelihood for image classification datasets.
Digital personas improve survey results for stable attributes but fail for subjective responses.
problem When can digital personas reliably approximate human survey findings?
method Using LISS panel, constructed personas from background variables and survey histories, tested against held-out post-cutoff answers.
result Digital personas improve alignment with human response distributions for stable attributes but fail for subjective responses.
Model shows how heterogeneity in strategies and risk tolerance affects financial market stability.
problem Understanding how heterogeneity impacts financial market dynamics.
method Agent-based model incorporating heterogeneous investment strategies and risk tolerance.
result Heterogeneity in strategies and risk tolerance suppresses price fluctuations.
Efficiently evaluate generative models at the prompt level using tensor factorization.
problem Fine-grained evaluations of generative models are costly and often misaligned with human judgment.
method Tensor factorization model that merges cheap autorater data with a small set of human gold-standard labels.
result The method provides accurate and tight confidence intervals for model performance.
Recognizing a hotel from an image of a hotel room is important for human trafficking investigations. Images directly link victims to places and can help verify where victims have been trafficked, and where their traffickers might move them or others in the future. Recognizing the hotel from images is challenging becaus…
ConCuR generates high-quality CUDA kernels with concise reasoning traces.
problem Scarce high-quality data for kernel generation.
method Developed a pipeline to generate and curate high-quality CUDA kernels with reasoning traces.
result Our model achieves significant improvements in KernelBench setup.
We present a novel extension of multi-output Gaussian processes for handling heterogeneous outputs. We assume that each output has its own likelihood function and use a vector-valued Gaussian process prior to jointly model the parameters in all likelihoods as latent functions. Our multi-output Gaussian process uses a c…
Framework learns stochastic dynamics from endpoint and intermediate distributions using soft energy constraints.
problem Learning stochastic dynamics from endpoint and intermediate distributional observations.
method Formulates generation as a McKean-Vlasov control problem with soft energy constraints, solving it through FBSDE.
result Model learns coherent stochastic trajectories matching prescribed marginal laws.
Machine learning categorizes mutual funds for better investment strategies.
problem Identifying similar mutual funds for diversified investment applications.
method Machine learning to learn and reproduce a well-regarded categorization system.
result Machine learning can learn and reproduce a categorization system that is widely regarded.
Develops a new tensor model for clustering with degree correction.
problem Clustering with unknown degree heterogeneity in multiway data.
method Degree-corrected tensor block model with estimation guarantees.
result Demonstrates an intrinsic statistical-to-computational gap for tensors of order three or greater.
Develops methods to correct bias in AI feedback for more accurate alignment.
problem Systematic bias in AI feedback compared to human labels.
method Two debiased alignment methods: DDPO and DIPO.
result Methods improve alignment efficiency and performance close to human-labeled data.