arXiv subjects

Yike Sun

Publications and source records attributed to Yike Sun.

5 recordsLinked to original sources

arXiv2026

ProofVerifier: A Scalable, Diversity-Driven Framework for Natural-Language Proof Verification

While large language models (LLMs) have achieved strong performance on mathematical problems with verifiable answers, many advanced problems are proof-based and require evaluating full proofs. However, training such verifiers requires diverse and trustworthy question-proof-check (QPC) examples at scale, which are scarce. To address this challenge, we develop a human-audited, LLM-assisted data pipeline that produces large-scale QPC triplets with limited human effort. By systematically varying problem sources, generation strategies, and generator models, the pipeline creates diverse problem-proof pairs spanning multiple difficulty levels, linguistic styles, and error types. We combine multi-LLM agreement with hierarchical human auditing to obtain accurate proof-correctness labels. Using these data, we train generative proof verifiers and introduce an auxiliary fluency filter together with balanced token weighting to stabilize binary-reward long-form verification RL. Experiments show that our verifier improves proof-judgment accuracy across different proof styles and provides useful guidance for test-time selection. Overall, our results provide a practical data and training framework for natural-language proof verification.

cs.CL↗

arXiv2026

CLIF: Concept-Level Influence Functions for Transparent Bottleneck Models

In recent years, the black-box nature of deep learning models has limited their application in high-stakes domains such as medical diagnosis and finance, where interpretability is essential. To address this, we propose a novel approach using influence functions to enhance interpretability in NLP models at both the sample and concept levels. Experiments on CEBaB and Yelp datasets show that influence functions effectively identify the most impactful training samples, both helpful and harmful, on model predictions. By adjusting the labels and weights of these samples, we demonstrate that model performance can be restored to baseline levels without retraining, confirming the value of influence functions for efficient data debugging. Furthermore, our concept-level analysis identifies key concepts within Concept Bottleneck Models (CBM) that significantly affect predictions. Modifying these concepts alters model behavior observably, providing clear insights into the decision process.

cs.CL↗

arXiv2026

LiteToken: Removing Intermediate Merge Residues From BPE Tokenizers

Tokenization is fundamental to how language models represent and process text, yet the behavior of widely used BPE tokenizers has received far less study than model architectures and training. In this paper, we investigate intermediate merge residues in BPE vocabularies: tokens that are frequent during merge learning so that retained in the final vocabulary, but are mostly further merged and rarely emitted when tokenizing the corpus during tokenizer usage. Such low-frequency tokens not only waste vocabulary capacity but also increase vulnerability to adversarial or atypical inputs. We present a systematic empirical characterization of this phenomenon across commonly used tokenizers and introduce LiteToken, a simple method for removing residue tokens. Because the affected tokens are rarely used, pretrained models can often accommodate the modified tokenizer without additional fine-tuning. Experiments show that LiteToken reduces token fragmentation, reduces parameters, and improves robustness to noisy or misspelled inputs, while preserving overall performance.

cs.CL↗

arXiv2025

Lateral Ventricular Brain-Computer Interface System with Lantern-Inspired Electrode for Stable Performance and Memory Decoding

We present a lateral ventricular brain-computer interface (LV-BCI) that deploys an expandable, flexible electrode into the lateral ventricle through a minimally invasive external ventricular drainage pathway. Inspired by the framework of traditional Chinese lanterns, the electrode expands uniformly within the ventricle and conforms to the ependymal wall. Compared with conventional subdural ECoG electrodes, the LV-BCI shows superior signal stability and immunocompatibility. Resting-state spectral analyses revealed a maximum effective bandwidth comparable to subdural ECoG. In evoked potential tests, the LV-BCI maintained a consistently higher signal-to-noise ratio over 112 days without the decline typically associated with scarring or other immune responses. Immunohistochemistry showed only a transient, early microglial activation after implantation, returning to control levels and remaining stable through 168 days. We further designed an "action-memory T-maze" task and developed a microstate sequence classifier (MSSC) to predict rats' turn decisions. The LV-BCI achieved prediction accuracy up to 98%, significantly outperforming subdural ECoG, indicating enhanced access to decision-related information from deep structures such as the hippocampus. These results establish the lateral ventricle as a viable route for neural signal acquisition. Using a lantern-inspired flexible electrode, we achieve long-term stable recordings and robust memory decision decoding from within the ventricular system, opening new directions for BCI technology and systems neuroscience.

q-bio.NC↗

arXiv2023

A Surgery-Detection Two-Dimensional Panorama of Signal Acquisition Technologies in Brain-Computer Interface

Brain-computer interface (BCI) technology is an interdisciplinary field that allows individuals to connect with the external world. The performance of BCI systems relies predominantly on the advancements of signal acquisition technology. This paper aims to present a comprehensive overview of signal acquisition technologies in BCI by examining research articles published in the past decade. Our review incorporates both clinician and engineer perspectives and presents a surgery-detection two-dimensional panorama of signal acquisition technologies in BCI. We classify the technologies into nine distinct categories, providing representative examples and emphasizing the significant challenges associated with each modality. Our review provides researchers and practitioners with a macroscopic understanding of BCI signal acquisition technologies and discuss the field's major issues today. Future development in BCI signal acquisition technology should prioritize the integration of diverse disciplines and perspectives. Striking a balance among signal quality, trauma, biocompatibility, and other relevant factors is crucial. This will promote the advancement of BCI technology, enhancing its efficiency, safety, and reliability, and ultimately contributing to a promising future for humanity.

cs.HC↗