# Actively Avoiding Nonsense in Generative Models

arXiv: https://arxiv.org/abs/1802.07229 · HTML: https://mlsanity.com/paper/1802.07229/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Feb 2018.

## Built on by (most cited first, 12 of 12)

- [Why Language Models Hallucinate](https://mlsanity.com/paper/2509.04664/) (2025)
- [Negative Data Augmentation](https://mlsanity.com/paper/2102.05113/) (2021)
- [Data Redaction from Pre-trained GANs](https://mlsanity.com/paper/2206.14389/) (2022)
- [Challenges of Large Language Models for Mental Health Counseling](https://mlsanity.com/paper/2311.13857/) (2023)
- [Evaluating Large Language Models as Virtual Annotators for Time-series Physical Sensing Data](https://mlsanity.com/paper/2403.01133/) (2024)
- [(Im)possibility of Automated Hallucination Detection in Large Language Models](https://mlsanity.com/paper/2504.17004/) (2025)
- [Constrained Generative Modeling with Manually Bridged Diffusion Models](https://mlsanity.com/paper/2502.20371/) (2025)
- [Don't be so negative! Score-based Generative Modeling with Oracle-assisted Guidance](https://mlsanity.com/paper/2307.16463/) (2023)
- [Novel Applications for VAE-based Anomaly Detection Systems](https://mlsanity.com/paper/2204.12577/) (2022)
- [Accelerating Radio Spectrum Regulation Workflows with Large Language Models (LLMs)](https://mlsanity.com/paper/2403.17819/) (2024)
- [Distribution Learning with Valid Outputs Beyond the Worst-Case](https://mlsanity.com/paper/2410.16253/) (2024)
- [Neither Stochastic Parroting nor AGI: LLMs Solve Tasks through Context-Directed Extrapolation from Training Data Priors](https://mlsanity.com/paper/2505.23323/) (2025)

## Abstract (arXiv)

A generative model may generate utter nonsense when it is fit to maximize the likelihood of observed data. This happens due to "model error," i.e., when the true data generating distribution does not fit within the class of generative models being learned. To address this, we propose a model of active distribution learning using a binary invalidity oracle that identifies some examples as clearly invalid, together with random positive examples sampled from the true distribution. The goal is to maximize the likelihood of the positive examples subject to the constraint of (almost) never generating examples labeled invalid by the oracle. Guarantees are agnostic compared to a class of probability distributions. We show that, while proper learning often requires exponentially many queries to the invalidity oracle, improper distribution learning can be done using polynomially many queries.