4arXiv cs.CL (Computation and Language)·
DWT-Fusion: Wavelet-based training-free framework for detecting LLM-generated text
Researchers introduce DWT-Fusion, a training-free method for detecting LLM-generated text that applies discrete wavelet transforms to token-level log-probability sequences from a proxy language model, capturing local and multiscale predictability patterns rather than global statistics. The framework evaluates four voting ensemble variants and is tested on HC3, M4, and MAGE benchmarks using GPT-Neo-2.7B, GPT-J-6B, Falcon-7B, and LLaMA-3-8B as proxy models. Best ensemble results achieve AUROC of 0.9919, 0.8477, and 0.7471 on the three benchmarks respectively. The approach is notable for requiring no supervised training while remaining interpretable.
Related guides (2)
Related events (8)
WaveDetect: Wavelet-based spectral fingerprinting for robust machine-generated text detection
WaveDetect is a new framework that reframes LLM-generated text detection as a signal processing problem, applying a differentiable Continuous Wavelet Transform to token probability sequences to extract 'spectral fingerprints' invisible in the time domain. The approach targets three known failure modes of existing detectors: adversarial perturbations, cross-domain shifts, and temporal model evolution. Evaluations on RAID, EvoBench, and Domain-Shift benchmarks claim state-of-the-art accuracy and robustness against sophisticated attacks and unseen LLMs.
Detecting LLM-Generated Text Using Classical Machine Learning Methods
A blog post explores using classical machine learning techniques (as opposed to neural approaches) to detect LLM-generated text. The post attracted notable Hacker News engagement with 168 points and 116 comments, suggesting practical interest from the community. LLM-generated text detection is an active area relevant to academic integrity, content moderation, and AI safety.
Trajectory Analysis of Masked Diffusion LMs for Graph-to-Text Generation with Lambda-Scaled Structural Decoding
This paper presents the first systematic study of masked diffusion language models (MDLMs) for graph-to-text generation, analyzing the order in which tokens are unmasked during iterative decoding. The authors find MDLMs naturally unmask entities first, then relational/function words, then structural tokens—a pattern disrupted by supervised fine-tuning, which prematurely anchors structural tokens and causes hallucination or omission. They propose lambda-scaled structural decoding, a training-free inference-time fix that recovers +9.4 BLEU-4, and introduce Graph-LLaDA, which integrates a Graph Transformer encoder into LLaDA's decoding process. Cross-dataset evaluation on the LAGRANGE benchmark shows prior baselines overfit to dataset-specific patterns while MDLM-based approaches generalize better.
Self-supervised WavLM + DTW framework for L2 pronunciation, rhythm, and intonation scoring
Researchers propose a text-free framework for second-language (L2) speech assessment using Dynamic Time Warping (DTW) over self-supervised WavLM representations, covering phonetic accuracy, rhythm, and intonation in English and Japanese. The DTW-based approach comparing learner speech to native templates exceeds human agreement on holistic phonetic scoring, and a novel warping-path method approaches human-level rhythm assessment. Intonation scoring, combining DTW over prosodic residuals with pitch and intensity features, shows more modest results. The method requires no labeled L2 data, making it applicable in low-resource settings.
Weave of Formal Thought: Sound-and-complete constrained decoding with learned latent syntax for code LLMs
The paper introduces Weave of Formal Thought (WoFT), a framework combining a formally sound-and-complete constrained decoder for code generation with a latent-variable fine-tuning method that teaches LLMs to interleave grammar non-terminals during generation. The constrained decoder extends generalized LR (GLR) parsing with speculative lexing to handle context-sensitive lexing and maximal-munch tokenization, addressing gaps in prior constrained-decoding work. A reweighted wake-sleep (RWS) fine-tuning objective on StarCoder2-3B achieves a 14.3% relative reduction in per-token cross-entropy over a text-only SFT baseline on Python, suggesting that explicit structural scaffolding recovers information lost in flat autoregressive training.
LLMSurgeon: Post-Hoc Auditing of LLM Pretraining Data Mixtures
LLMSurgeon formalizes Data Mixture Surgery (DMS), a framework for estimating the domain-level distribution of an LLM's pretraining corpus using only generated text from the target model. The method casts DMS as an inverse problem under the label-shift assumption, using a calibrated soft confusion matrix to correct domain confusion and recover the latent mixture prior. The authors also introduce LLMScan, a verifiable evaluation suite built from open-source LLMs with known pretraining mixtures, on which LLMSurgeon demonstrates high-fidelity recovery of domain compositions without access to training data.
LLMs outperform traditional methods on single and multi-truth data fusion tasks
A new arXiv preprint investigates using LLMs for data fusion (truth discovery) over tabular data, covering both single-truth and multi-truth scenarios. The authors evaluate domain-dependent, domain-independent, zero-shot, and one-shot prompting strategies across three benchmark datasets. LLM-based approaches outperform traditional unsupervised methods including DART and LTM on all datasets, with code released publicly.
Distill to Detect (D2D): surfacing stealth preferential biases in LLMs via cartridge distillation
Researchers introduce Distill to Detect (D2D), a method for exposing hidden preferential biases in language models by distilling the distributional shift between a suspected model and its base into a KV-cache prefix adapter (cartridge), amplifying the bias signal into generated text. The work addresses a fundamental asymmetry in bias detection: stealth biases introduced anywhere in a model's supply chain can remain invisible to text-based inspection, weight analysis, or representation probing. D2D is grounded in a theoretical framework based on Fisher-weighted projection of logit distribution shifts and is validated across multiple bias types. The technique offers a practical auditing tool for detecting hidden behaviors in deployed LLMs.