Gregory Kang Ruey Lau

Kernelized bandits, also known as Bayesian optimization (BO), has been a prevalent method for optimizing complicated black-box reward functions. Various BO algorithms have been theoretically shown to enjoy upper bounds on their cumulative regret which are sub-linear in the number T of iterations, and a regret lower bound of Ω(sqrtT) has been derived which represents the unavoidable regrets for any classical BO algorithm. Recent works on quantum bandits have shown that with the aid of quantum computing, it is possible to achieve tighter regret upper bounds better than their corresponding classical lower bounds. However, these works are restricted to either multi-armed or linear bandits, and are hence not able to solve sophisticated real-world problems with non-linear reward functions. To this end, we introduce the quantum-Gaussian process-upper confidence bound (Q-GP-UCB) algorithm. To the best of our knowledge, our Q-GP-UCB is the first BO algorithm able to achieve a regret upper bound of O(polylog T), which is significantly smaller than its regret lower bound of Ω(sqrtT) in the classical setting. Moreover, thanks to our novel analysis of the confidence ellipsoid, our Q-GP-UCB with the linear kernel achieves a smaller regret than the quantum linear UCB algorithm from the previous work. We use simulations, as well as an experiment using a real quantum computer, to verify that the theoretical quantum speedup achieved by our Q-GP-UCB is also potentially relevant in practice.

Physics-Informed Neural Networks (PINNs), which incorporate PDEs as soft constraints, train with a composite loss function that contains multiple training point types: different types of \textitcollocation points chosen during training to enforce each PDE and initial/boundary conditions, and \textitexperimental points which are usually costly to obtain via experiments or simulations. Training PINNs using this loss function is challenging as it typically requires selecting large numbers of points of different types, each with different training dynamics. Unlike past works that focused on the selection of either collocation or experimental points, this work introduces \textscPINN Adaptive ColLocation and Experimental points selection (\alg), the first algorithm that \emphjointly optimizes the selection of all training point types, while automatically adjusting the proportion of collocation point types as training progresses. \alg uses information on the interaction among training point types, which had not been considered before, based on an analysis of PINN training dynamics via the Neural Tangent Kernel (NTK). We theoretically show that the criterion used by \alg is related to the PINN generalization error, and empirically demonstrate that \alg is able to outperform existing point selection methods for forward, inverse, and transfer learning problems.

In many science and engineering settings, system dynamics are characterized by governing partial differential equations (PDEs), and a major challenge is to solve inverse problems (IPs) where unknown PDE parameters are inferred based on observational data gathered under limited budget. Due to the high costs of setting up and running experiments, experimental design (ED) is often done with the help of PDE simulations to optimize for the most informative design parameters (e.g., sensor placements) to solve such IPs, prior to actual data collection. This process of optimizing design parameters is especially critical when the budget and other practical constraints make it infeasible to adjust the design parameters between trials during the experiments. However, existing experimental design (ED) methods tend to require sequential and frequent design parameter adjustments between trials. Furthermore, they also have significant computational bottlenecks due to the need for complex numerical simulations for PDEs, and do not exploit the advantages provided by physics informed neural networks (PINNs) in solving IPs for PDE-governed systems, such as its meshless solutions, differentiability, and amortized training. This work presents Physics-Informed Experimental Design (PIED), the first ED framework that makes use of PINNs in a fully differentiable architecture to perform continuous optimization of design parameters for IPs for one-shot deployments. PIED overcomes existing methods’ computational bottlenecks through parallelized computation and meta-learning of PINN parameter initialization, and proposes novel methods to effectively take into account PINN training dynamics in optimizing the ED parameters. Through experiments based on noisy simulated data and even real world experimental data, we empirically show that given limited observation budget, PIED significantly outperforms existing ED methods in solving IPs, including for challenging settings where the inverse parameters are unknown functions rather than just finite-dimensional.

Protecting intellectual property (IP) of text such as articles and code is increasingly important, especially as sophisticated attacks become possible, such as paraphrasing by large language models (LLMs) or even unauthorized training of LLMs on copyrighted text to infringe such IP. However, existing text watermarking methods are not robust enough against such attacks nor scalable to millions of users for practical implementation. In this paper, we propose Waterfall, the first training-free framework for robust and scalable text watermarking applicable across multiple text types (e.g., articles, code) and languages supportable by LLMs, for general text and LLM data provenance. Waterfall comprises several key innovations, such as being the first to use LLM as paraphrasers for watermarking along with a novel combination of techniques that are surprisingly effective in achieving robust verifiability and scalability. We empirically demonstrate that Waterfall achieves significantly better scalability, robust verifiability, and computational efficiency compared to SOTA article-text watermarking methods, and also showed how it could be directly applied to the watermarking of code.

Large Language Models (LLMs), particularly smaller variants, still struggle with complex reasoning tasks. While inference-time prompting can guide reasoning, existing methods often rely on sequential queries. Ensemble approaches offer a promising path to performance gains, especially given recent batch inference speed-ups. This work introduces DIPPER, a novel, training-free framework that transforms a single LLM into an effective inference-time ensemble. By feeding the model an optimized and diverse set of prompts in parallel, DIPPER elicits varied reasoning paths, leading to performance gains. We empirically demonstrate significant improvements on mathematical reasoning benchmarks, such as MATH, where a DIPPER ensemble of three Qwen2-MATH-1.5B instances (via parallel prompting of a single model) outperforms a larger Qwen2-MATH-7B model.

Large language model (LLM) unlearning is critical in real-world applications where it is necessary to efficiently remove the influence of private, copyrighted, or harmful data from some users. Existing utility-centric unlearning metrics (based on model utility) may fail to accurately evaluate the extent of unlearning in realistic settings such as when the forget and retain sets have semantically similar content and/or retraining the model from scratch on the retain set is impractical. This paper presents the first data-centric unlearning metric for LLMs called WaterDrum that exploits robust text watermarking to overcome these limitations. We introduce new benchmark datasets (with different levels of data similarity) for LLM unlearning that can be used to rigorously evaluate unlearning algorithms via WaterDrum.

The performance of an LLM depends heavily on the relevance of its training data to the downstream evaluation task. However, in practice, we do not have fine-grained knowledge of the data in the evaluation task (e.g., conversations between an LLM and a user are end-to-end encrypted). Hence, it is unclear what data is relevant for fine-tuning the LLM. Instead, we can only deploy the LLM on the unseen task to gather multiple rounds of coarse, noisy feedback on how well the model performs (e.g., user ratings). Our paper presents DUET, a novel global-to-local algorithm that optimizes training data mixtures by interleaving data selection with Bayesian optimization to exploit coarse and noisy feedback from a downstream evaluation task. DUET is flexible enough to incorporate different data selection methods, each with different performance-compute tradeoffs. By analyzing DUET’s cumulative regret, we theoretically show that DUET converges to the optimal training data mixture even without any fine-grained data information from an unseen task. Finally, our experiments across a variety of language tasks demonstrate that DUET attains substantial performance improvement over existing data selection and mixing methods in the unseen-task setting. Our anonymized code can be found at https://github.com/pmsdapfmbf/DUET.

This position paper proposes a data-centric viewpoint of AI research, focusing on large language models (LLMs). We start by making the key observation that data is instrumental in the developmental (e.g., pretraining and fine-tuning) and inferential stages (e.g., in-context learning) of LLMs, and yet it receives disproportionally low attention from the research community. We identify four specific scenarios centered around data, covering data-centric benchmarks and data curation, data attribution, knowledge transfer, and inference contextualization. In each scenario, we underscore the importance of data, highlight promising research directions, and articulate the potential impacts on the research community and, where applicable, the society as a whole. For instance, we advocate for a suite of data-centric benchmarks tailored to the scale and complexity of data for LLMs. These benchmarks can be used to develop new data curation methods and document research efforts and results, which can help promote openness and transparency in AI and LLM research.

Large Language Models (LLMs) are large-scale pretrained models that have achieved remarkable success across diverse domains. These successes have been driven by unprecedented complexity and scale in both data and computations. However, due to the high costs of training such models, brute-force trial-and-error approaches to improve LLMs are not feasible. Inspired by the success of inverse problems in uncovering fundamental scientific laws, this position paper advocates that inverse problems can also be used to efficiently uncover scaling laws that guide the building of LLMs to achieve a desirable performance with significantly better cost-effectiveness.

In this paper, we propose the first training-free framework for robust and scalable text watermarking applicable across multiple text types (e.g., articles, code) and languages, for general as well as LLM text training data provenance. We highlight perspectives on text IP protection, such as using LLMs to enable better IP protection rather than viewing them as just sources of IP infringement, not relying on just major LLM providers, and the benefits of having a general framework that can be easily adapted to defend against new threats.

Despite their capabilities, Multimodal Large Language Models (MLLMs) may produce plausible but erroneous outputs, hindering reliable deployment. Accurate uncertainty metrics could enable escalation of unreliable queries to human experts or larger models for improved performance. However, existing uncertainty metrics have practical constraints, such as being designed only for specific modalities, reliant on external tools, or computationally expensive. We introduce UMPIRE, a training-free uncertainty quantification framework for MLLMs that works efficiently across various input and output modalities without external tools, relying only on the models’ own internal modality features. UMPIRE computes the incoherence-adjusted semantic volume of sampled MLLM responses for a given task instance, effectively capturing both the global semantic diversity of samples and the local incoherence of responses based on internal model confidence. We propose uncertainty desiderata for MLLMs and provide theoretical analysis motivating UMPIRE’s design. Extensive experiments show that UMPIRE consistently outperforms baseline metrics in error detection and uncertainty calibration across image, audio, and video-text benchmarks, including adversarial and out-of-distribution settings. We also demonstrate UMPIRE’s generalization to non-text output tasks, including image and audio generation.

Discovering scientific laws or interpretable symbolic equations from data rapidly is important in many setting, such as decision-making in time-sensitive high-stake scenarios or applications involving interactive or iterative experimentation such as in scientific or machine learning workflows. However, existing methods, generally known as symbolic regression (SR), typically require long computational time to achieve good performance and have to run from scratch for each dataset. Recent methods that use pre-training SR foundation models for faster inference also suffer from performance limitations and require large training datasets. In this work, we propose README, a framework for rapid equation discovery that can generate performant, interpretable equations from limited, noisy data in just a few seconds, and requires significantly less training data compared to past SR foundation model approaches. We achieve this by being the first to (1) work with image representations of datasets to efficiently capture their key properties, (2) combine the capabilities of open-sourced pre-trained text and image encoders to produce an informative SR embedding space, and (3) develop a novel Grey Wolf Optimizer with Bayesian Optimization (GWOBO) algorithm to rapidly optimize for the best symbolic expression within seconds. We empirically show that README outperforms benchmarks on a wide range of realistic datasets, including real experimental data from various domains and noisy video-extracted dynamics.

While Multimodal Large Language Models (MLLMs) exhibit strong reasoning on text inputs, they often fail on semantically equivalent visual inputs. By rendering text problems as images, we isolate this failure and identify a reasoning-access gap: models correctly perceive visual content but fail to route that content into the latent reasoning mechanisms used for text-based tasks. To address this, we propose TIGER (Text-to-Image Gap-targeted Training for Enhanced Reasoning). TIGER automatically transforms text-only corpora into multimodal training data by mining modality counterfactuals, instances where a model succeeds on text but fails on the equivalent image, providing targeted supervision without manually curated datasets. Implemented via image-conditioned Group Relative Policy Optimization (GRPO), TIGER consistently narrows the modality gap and improves visual reasoning on benchmarks like MathVerse and EMMA. We further show that even RLVR-based models exhibit modality-dependent reasoning gaps, and that TIGER effectively reduces them. Furthermore, activation analyses reveal that TIGER helps visual representations better engage reasoning-relevant subspaces within the language backbone. Our results emphasize that robust multimodal reasoning requires reliable visual access to existing reasoning machinery, moving beyond better perception.

Data provenance aims to determine whether and how a data source has influenced a downstream LLM. Despite growing interest in data provenance research, current methods tend to specialize on specific settings and suffer from fragmented evaluation standards. To address this, we introduce WATERSHED, a unified benchmark and toolkit for end-to-end provenance evaluation. WATERSHED structures data provenance into stage-wise tests spanning data preparation, LLM training, black-box auditing, and downstream applications such as membership audit, multi-owner source attribution, and unlearning verification. We evaluate existing provenance methods such as watermarking and membership inference attacks on WATERSHED, across a wide range of datasets, model families and attacks. Our results confirm that methods vary in effectiveness across different stages and tasks. By providing a unified framework and exposing these failure modes, WATERSHED establishes a rigorous basis for evaluating data provenance methods.