6 papers
Select or Rewrite: Oracle-Guided Fine-Tuning for Generative ASR Post-Editing from N-Best Hypotheses
Shiran Liu, Gaohui Xu, Sergio Burdisso +3
Large language models can post-edit ASR by using linguistic context to compare N-best hypotheses and rewrite the transcript when needed, often improving WER without retraining the…
Counting and Covering in Nearest-Neighbour Representations of Boolean Functions
Martin Anthony
We study the number of prototypes needed to represent Boolean functions by nearest-neighbour classification. There are two distinct settings: the prototypes may be arbitrary points…
Finite-Sample Coverage Audits for High-Recall Candidate Generation: Certification and Learning-Theoretic Design
Martin Anthony, Kaveh Salehzadeh Nobari
An initial high-recall stage in an empirical pipeline decides which items pass to later review, labelling, or modelling, and relevant items it misses are lost to every subsequent s…
Chamber geometry and specification numbers of Boolean threshold functions
Martin Anthony
The specification number of a Boolean threshold function on variables is the least number of points whose -values determine uniquely among all threshold fun…
Generalised Eigenvalue Geometry of Semantic Adversarial Attacks
Martin Anthony, Kaveh Salehzadeh Nobari
Recent empirical work shows that semantically equivalent paraphrases can fool financial sentiment classifiers: although a paraphrase remains close to the original under a strong re…
Quadratization of Symmetric Pseudo-Boolean Functions
Martin Anthony, Endre Boros, Yves Crama +1
A pseudo-Boolean function is a real-valued function of binary variables; that is, a mapping from to . For a pseudo-Boolean…