SearcharxivSearch

arXiv subjects

Gaofei Wu

Publications and source records attributed to Gaofei Wu.

9 recordsLinked to original sources

KSHSeek: Data-Driven Approaches to Mitigating and Detecting Knowledge-Shortcut Hallucinations in Generative Models

The emergence of large language models (LLMs) has significantly advanced the development of natural language processing (NLP), especially in text generation tasks like question answering. However, model hallucinations remain a major challenge in natural language generation (NLG) tasks due to their complex causes. We systematically expand on the causes of factual hallucinations from the perspective of knowledge shortcuts, analyzing hallucinations arising from correct and defect-free data and demonstrating that knowledge-shortcut hallucinations are prevalent in generative models. To mitigate this issue, we propose a high similarity pruning algorithm at the data preprocessing level to reduce spurious correlations in the data. Additionally, we design a specific detection method for knowledge-shortcut hallucinations to evaluate the effectiveness of our mitigation strategy. Experimental results show that our approach effectively reduces knowledge-shortcut hallucinations, particularly in fine-tuning tasks, without negatively impacting model performance in question answering. This work introduces a new paradigm for mitigating specific hallucination issues in generative models, enhancing their robustness and reliability in real-world applications.

cs.CL

Several new classes of optimal ternary cyclic codes with two or three zeros

Cyclic codes are a subclass of linear codes and have wide applications in data storage systems, communication systems and consumer electronics due to their efficient encoding and decoding algorithms. Let $\alpha $ be a generator of $\mathbb{F}_{3^m}^*$, where $m$ is a positive integer. Denote by $\mathcal{C}_{(i_1,i_2,\cdots, i_t)}$ the cyclic code with generator polynomial $m_{\alpha^{i_1}}(x)m_{\alpha^{i_2}}(x)\cdots m_{\alpha^{i_t}}(x)$, where ${{m}_{\alpha^{i}}}(x)$ is the minimal polynomial of ${{\alpha }^{i}}$ over ${{\mathbb{F}}_{3}}$. In this paper, by analyzing the solutions of certain equations over finite fields, we present four classes of optimal ternary cyclic codes $\mathcal{C}_{(0,1,e)}$ and $\mathcal{C}_{(1,e,s)}$ with parameters $[3^m-1,3^m-\frac{3m}{2}-2,4]$, where $s=\frac{3^m-1}{2}$. In addition, by determining the solutions of certain equations and analyzing the irreducible factors of certain polynomials over $\mathbb{F}_{3^m}$, we present four classes of optimal ternary cyclic codes $\mathcal{C}_{(2,e)}$ and $\mathcal{C}_{(1,e)}$ with parameters $[3^m-1,3^m-2m-1,4]$. We show that our new optimal cyclic codes are inequivalent to the known ones.

cs.IT

Catch the Butterfly: Peeking into the Terms and Conflicts among SPDX Licenses

The widespread adoption of third-party libraries (TPLs) in software development has accelerated the creation of modern software. However, this convenience comes with potential legal risks. Developers may inadvertently violate the licenses of TPLs, leading to legal issues. While existing studies have explored software licenses and potential incompatibilities, these studies often focus on a limited set of licenses or rely on low-quality license data, which may affect their conclusions. To address this gap, there is a need for a high-quality license dataset that encompasses a broad range of mainstream licenses to help developers navigate the complex landscape of software licenses, avoid potential legal pitfalls, and guide solutions for managing license compliance and compatibility in software development. To this end, we conduct the first work to understand the mainstream software licenses based on term granularity and obtain a high-quality dataset of 453 SPDX licenses with well-labeled terms and conflicts. Specifically, we first conduct a differential analysis of the mainstream platforms to understand the terms and attitudes of each license. Next, we propose a standardized set of license terms to capture and label existing mainstream licenses with high quality. Moreover, we include copyleft conflicts and conclude the three major types of license conflicts among the 453 SPDX licenses. Based on these, we carry out two empirical studies to reveal the concerns and threats from the perspectives of both licensors and licensees. One study provides an in-depth analysis of the similarities, differences, and conflicts among SPDX licenses, revisits the usage and conflicts of licenses in the NPM ecosystem, and draws conclusions that differ from previous work. Our studies reveal some insightful findings and disclose relevant analytical data, which set the stage for further research.

cs.SE

Several classes of optimal $p$-ary cyclic codes with minimal distance four

Cyclic codes are a subclass of linear codes and have wide applications in data storage systems, communication systems and consumer electronics due to their efficient encoding and decoding algorithms. Let $p\ge 5$ be an odd prime and $m$ be a positive integer. Let $\mathcal{C}_{(1,e,s)}$ denote the $p$-ary cyclic code with three nonzeros $\alpha$, $\alpha^e$, and $\alpha^s$, where $\alpha $ is a generator of ${\mathbb F}_{p^m}^*$, $s=\frac{p^m-1}{2}$, and $2\le e\le p^m-2$. In this paper, we present four classes of optimal $p$-ary cyclic codes $\mathcal{C}_{(1,e,s)}$ with parameters $[p^m-1,p^m-2m-2,4]$ by analyzing the solutions of certain polynomials over finite fields. Some previous results about optimal quinary cyclic codes with parameters $[5^m-1,5^m-2m-2,4]$ are special cases of our constructions. In addition, by analyzing the irreducible factors of certain polynomials over ${\mathbb F}_{5^m}$, we present two classes of optimal quinary cyclic codes $\mathcal{C}_{(1,e,s)}$.

cs.IT

A New Method to Construct Gloay Complementary Set by Paraunitary Matrices and Hadamard Matrices

Golay complementary sequences have been put a high value on the applications in orthogonal frequency-division multiplexing (OFDM) systems since its good peak-to-mean envelope power ratio(PMEPR) properties. However, with the increase of the code length, the code rate of the standard Golay sequences suffer a dramatic decline. Even though a lot of efforts have been paid to solve the code rate problem for OFDM application, how to construct large classes of sequences with low PMEPR is still difficult and open now. In this paper, we propose a new method to construct $q$-ary Golay complementary set of size $N$ and length $N^n$ by $N\times N$ Hadamard Matrices where $n$ is arbitrary and $N$ is a power of 2. Every item of the constructed sequences can be presented as the product of the specific entries of the Hadamard Matrices. The previous works in \cite{BudIT} can be regarded as a special case of the constructions in this paper and we also obtained new quaternary Golay sets never reported in the literature.

cs.IT

Several Classes of Permutation Trinomials over $\mathbb F_{5^n}$ From Niho Exponents

The construction of permutation trinomials over finite fields attracts people's interest recently due to their simple form and some additional properties. Motivated by some results on the construction of permutation trinomials with Niho exponents, by constructing some new fractional polynomials that permute the set of the $(q+1)$-th roots of unity in $\mathbb F_{q^2}$, we present several classes of permutation trinomials with Niho exponents over $\mathbb F_{q^2}$, where $q=5^k$.

cs.IT

Several Classes of Negabent Functions over Finite Fields

Negabent functions as a class of generalized bent functions have attracted a lot of attention recently due to their applications in cryptography and coding theory. In this paper, we consider the constructions of negabent functions over finite fields. First, by using the compositional inverses of certain binomial and trinomial permutations, we present several classes of negabent functions of the form $f(x)=\Tr_1^n(λx^{2^k+1})+\Tr_1^n(ux)\Tr_1^n(vx)$, where $λ\in \F_{2^n}$, $2\leq k\leq n-1$, $(u,v)\in \F^*_{2^n}\times \F^*_{2^n}$, and $\Tr_1^n(\cdot)$ is the trace function from $\F_{2^n}$ to $\F_{2}$. Second, by using Kloosterman sum, we prove that the condition for the cubic monomials given by Zhou and Qu (Cryptogr. Commun., to appear, DOI 10.1007/s12095-015-0167-0.) to be negabent is also necessary. In addition, a conjecture on negabent monomials whose exponents are of Niho type is given.

cs.IT

More Classes of Complete Permutation Polynomials over $\F_q$

In this paper, by using a powerful criterion for permutation polynomials given by Zieve, we give several classes of complete permutation monomials over $\F_{q^r}$. In addition, we present a class of complete permutation multinomials, which is a generalization of recent work.

cs.IT

A complementary construction using mutually unbiased bases

We present a construction for complementary pairs of arrays that exploits a set of mutually-unbiased bases, and enumerate these arrays as well as the corresponding set of complementary sequences obtained from the arrays by projection. We also sketch an algorithm to uniquely generate these sequences. The pairwise squared inner-product of members of the sequence set is shown to be $\frac{1}{2}$. Moreover, a subset of the set can be viewed as a codebook that asymptotically achieves $\sqrt{\frac{3}{2}}$ times the Welch bound.

cs.IT