Real or Fake? Learning to Discriminate Machine from Human Generated Text

Abstract

Energy-based models (EBMs), a.k.a. un-normalized models, have had recent successes in continuous spaces. However, they have not been successfully applied to model text sequences. While decreasing the energy at training samples is straightforward, mining (negative) samples where the energy should be increased is difficult. In part, this is because standard gradient-based methods are not readily applicable when the input is high-dimensional and discrete. Here, we side-step this issue by generating negatives using pre-trained auto-regressive language models. The EBM then works in the residual of the language model; and is trained to discriminate real text from text generated by the auto-regressive models. We investigate the generalization ability of residual EBMs, a pre-requisite for using them in other applications. We extensively analyze generalization for the task of classifying whether an input is machine or human generated, a natural task given the training loss and how we mine negatives. Overall, we observe that EBMs can generalize remarkably well to changes in the architecture of the generators producing negatives. However, EBMs exhibit more sensitivity to the training set used by such generators.

References in corpus (9)

• Adam: A Method for Stochastic Optimization
• Sequence to Sequence Learning with Neural Networks
• Natural Language Processing (almost) from Scratch
• On the difficulty of training Recurrent Neural Networks
• SGDR: Stochastic Gradient Descent with Warm Restarts
• Pointer Sentinel Mixture Models
• Fast Decoding in Sequence Models using Discrete Latent Variables
• Implicit Generation and Generalization in Energy-Based Models
• Insertion-based Decoding with automatically Inferred Generation Order

Cited by in corpus (12)

• Release Strategies and the Social Impacts of Language Models
• TweepFake: about Detecting Deepfake Tweets
• Artificial Text Detection via Examining the Topology of Attention Maps
• Residual Energy-Based Models for Text Generation
• Style Example-Guided Text Generation using Generative Adversarial Transformers
• Cascaded Text Generation with Markov Transformers
• Analyzing the Forgetting Problem in the Pretrain-Finetuning of Dialogue Response Models
• Detecting Individual Decision-Making Style: Exploring Behavioral Stylometry in Chess
• Compositional Visual Generation and Inference with Energy Based Models
• All-in-One Image-Grounded Conversational Agents
• Neural Deepfake Detection with Factual Structure of Text
• Reverse Engineering Configurations of Neural Text Generation Models