Search

Large language models (LLMs) such as ChatGPT have exhibited remarkable performance in generating human-like texts. However, machine-generated texts (MGTs) may carry critical risks, such as plagiarism issues, misleading information, or hallucination issues. Therefore, it is very urgent and important to detect MGTs in many situations. Unfortunately, it is challenging to distinguish MGTs and human-written texts because the distributional discrepancy between them is often very subtle due to the remarkable performance of LLMs. In this paper, we seek to exploit \textit{maximum mean discrepancy} (MMD) to address this issue in the sense that MMD can well identify distributional discrepancies. However, directly training a detector with MMD using diverse MGTs will incur a significantly increased variance of MMD since MGTs may contain \textit{multiple text populations} due to various LLMs. This will severely impair MMD's ability to measure the difference between two samples. To tackle this, we propose a novel \textit{multi-population} aware optimization method for MMD called MMD-MP, which can \textit{avoid variance increases} and thus improve the stability to measure the distributional discrepancy. Relying on MMD-MP, we develop two methods for paragraph-based and sentence-based detection, respectively. Extensive experiments on various LLMs, \eg, GPT2 and ChatGPT, show superior detection performance of our MMD-MP. The source code is available at \url{https://github.com/ZSHsh98/MMD-MP}., Comment: Accepted at ICLR 2024

AI-powered programming language generation (PLG) models have gained increasing attention due to their ability to generate source code of programs in a few seconds with a plain program description. Despite their remarkable performance, many concerns are raised over the potential risks of their development and deployment, such as legal issues of copyright infringement induced by training usage of licensed code, and malicious consequences due to the unregulated use of these models. In this paper, we present the first-of-its-kind study to systematically investigate the accountability of PLG models from the perspectives of both model development and deployment. In particular, we develop a holistic framework not only to audit the training data usage of PLG models, but also to identify neural code generated by PLG models as well as determine its attribution to a source model. To this end, we propose using membership inference to audit whether a code snippet used is in the PLG model's training data. In addition, we propose a learning-based method to distinguish between human-written code and neural code. In neural code attribution, through both empirical and theoretical analysis, we show that it is impossible to reliably attribute the generation of one code snippet to one model. We then propose two feasible alternative methods: one is to attribute one neural code snippet to one of the candidate PLG models, and the other is to verify whether a set of neural code snippets can be attributed to a given PLG model. The proposed framework thoroughly examines the accountability of PLG models which are verified by extensive experiments. The implementations of our proposed framework are also encapsulated into a new artifact, named CodeForensic, to foster further research.

The emerging optical multiplexing within nanoscale shows super-capacity in encoding information by using the time-domain fingerprints from uniform nanoparticles. However, the optical diffraction limit compromises the decoding throughput and accuracy of the nanoparticles during wide-field imaging. This, in turn, challenges the quality of nanoparticles to afford the modulated excitation condition, and further to retain the multiplexed optical fingerprints for super-resolution multiplexing. Here we report a tailor-made time-domain super-resolution method with the lifetime-engineered upconversion nanoparticles for multiplexing. We demonstrate that the nanoparticles are bright, uniform, and stable under structured illumination, which supports a lateral resolution of 186 nm, less than 1/4th of the excitation wavelength. We further develop a deep learning algorithm to coordinate with super-resolution images for more accurate decoding compared to a numeric algorithm. We demonstrate a three-channel sub-diffraction-limit imaging-based optical multiplexing with decoding accuracies above 93% for each channel, and larger than 60% accuracies for potential seven-channel multiplexing. The improved resolution provides high throughput by resolving the particles within the optical limit, which enables higher multiplexing capacity in space. This time-domain super-resolution multiplexing opens a new horizon for handling the growing amount of information content, diseases source, and security risk in modern society, Comment: 10 pages, 5 figures

The control in optical uniformity of single nanoparticles and tuning their diversity in orthogonal dimensions, dot to dot, holds the key to unlock nanoscience and applications. Here we report that the time-domain emissive profile from single upconversion nanoparticle, including the rising, decay and peak moment of the excited state population (T2 profile), can be arbitrarily tuned by upconversion schemes, including interfacial energy migration, concentration dependency, energy transfer, and isolation of surface quenchers. This allows us to significantly increase the coding capacity at the nanoscale. We further implement both time-resolved wide-field imaging and deep-learning techniques to decode these fingerprints, showing high accuracies at high throughput. These high-dimensional optical fingerprints provide a new horizon for applications spanning from sub-diffraction-limit data storage, security inks, to high-throughput single-molecule digital assays and super-resolution imaging.