no code implementations • ICML 2020 • Dmitrii Kharkovskii, Zhongxiang Dai, Bryan Kian Hsiang Low
This paper presents the outsourced-Gaussian process-upper confidence bound (O-GP-UCB) algorithm, which is the first algorithm for privacy-preserving Bayesian optimization (BO) in the outsourced setting with a provable performance guarantee.
no code implementations • ICML 2020 • Nghia Hoang, Thanh Lam, Bryan Kian Hsiang Low, Patrick Jaillet
The task-agnostic prototypes can then be reintegrated to generate a new model that solves a new task encoded with a different prototype distribution.
3 code implementations • 11 Apr 2024 • Gregory Kang Ruey Lau, Apivich Hemachandra, See-Kiong Ng, Bryan Kian Hsiang Low
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 collocation points chosen during training to enforce each PDE and initial/boundary conditions, and experimental points which are usually costly to obtain via experiments or simulations.
no code implementations • NeurIPS 2023 • Rachael Hwee Ling Sim, Yehong Zhang, Trong Nghia Hoang, Xinyi Xu, Bryan Kian Hsiang Low, Patrick Jaillet
Finally, the mediator rewards each party with different posterior samples of the model parameters.
1 code implementation • 12 Mar 2024 • Zhenfeng He, Yao Shu, Zhongxiang Dai, Bryan Kian Hsiang Low
Nevertheless, the estimation ability of these metrics typically varies across different tasks, making it challenging to achieve robust and consistently good search performance on diverse tasks with only a single training-free metric.
no code implementations • 5 Mar 2024 • Wenyang Hu, Yao Shu, Zongmin Yu, Zhaoxuan Wu, Xiangqiang Lin, Zhongxiang Dai, See-Kiong Ng, Bryan Kian Hsiang Low
Existing methodologies usually prioritize a global optimization for finding the global optimum, which however will perform poorly in certain tasks.
no code implementations • 4 Feb 2024 • Zhuanghua Liu, Bryan Kian Hsiang Low
However, the convergence rate of the PMGT-SVRG algorithm has a linear dependency on the condition number, which is undesirable for the ill-conditioned problem.
no code implementations • 4 Feb 2024 • Zhuanghua Liu, Luo Luo, Bryan Kian Hsiang Low
The recently proposed incremental quasi-Newton method is based on BFGS update and achieves a local superlinear convergence rate that is dependent on the condition number of the problem.
1 code implementation • 26 Jan 2024 • Rui Qiao, Bryan Kian Hsiang Low
Despite the rapid development of machine learning algorithms for domain generalization (DG), there is no clear empirical evidence that the existing DG algorithms outperform the classic empirical risk minimization (ERM) across standard benchmarks.
1 code implementation • 18 Dec 2023 • Xiao Tian, Rachael Hwee Ling Sim, Jue Fan, Bryan Kian Hsiang Low
Data valuation is concerned with determining a fair valuation of data from data sources to compensate them or to identify training examples that are the most or least useful for predictions.
1 code implementation • 2 Oct 2023 • Xiaoqiang Lin, Zhaoxuan Wu, Zhongxiang Dai, Wenyang Hu, Yao Shu, See-Kiong Ng, Patrick Jaillet, Bryan Kian Hsiang Low
We perform instruction optimization for ChatGPT and use extensive experiments to show that our INSTINCT consistently outperforms the existing methods in different tasks, such as in various instruction induction tasks and the task of improving the zero-shot chain-of-thought instruction.
no code implementations • 1 Oct 2023 • Jingtan Wang, Xinyang Lu, Zitong Zhao, Zhongxiang Dai, Chuan-Sheng Foo, See-Kiong Ng, Bryan Kian Hsiang Low
The impressive performances of large language models (LLMs) and their immense potential for commercialization have given rise to serious concerns over the intellectual property (IP) of their training data.
1 code implementation • 8 Aug 2023 • Yao Shu, Xiaoqiang Lin, Zhongxiang Dai, Bryan Kian Hsiang Low
To this end, we (a) introduce trajectory-informed gradient surrogates which is able to use the history of function queries during optimization for accurate and query-efficient gradient estimation, and (b) develop the technique of adaptive gradient correction using these gradient surrogates to mitigate the aforementioned disparity.
no code implementations • 1 Aug 2023 • Mohit Rajpal, Lac Gia Tran, Yehong Zhang, Bryan Kian Hsiang Low
Derivative-free approaches such as Bayesian Optimization mitigate the dependency on the quality of gradient feedback, but are known to scale poorly in the high-dimension setting of complex decision making systems.
1 code implementation • 9 Jun 2023 • Xiaoqiang Lin, Xinyi Xu, See-Kiong Ng, Chuan-Sheng Foo, Bryan Kian Hsiang Low
In collaborative learning with streaming data, nodes (e. g., organizations) jointly and continuously learn a machine learning (ML) model by sharing the latest model updates computed from their latest streaming data.
1 code implementation • 7 Jun 2023 • Apivich Hemachandra, Zhongxiang Dai, Jasraj Singh, See-Kiong Ng, Bryan Kian Hsiang Low
To this end, we introduce our expected variance with Gaussian processes (EV-GP) criterion for neural active learning, which is theoretically guaranteed to select data points which lead to trained NNs with both (a) good predictive performances and (b) initialization robustness.
1 code implementation • 23 May 2023 • Tiedong Liu, Bryan Kian Hsiang Low
We introduce Goat, a fine-tuned LLaMA model that significantly outperforms GPT-4 on a range of arithmetic tasks.
no code implementations • 9 May 2023 • Zhang Ze Yu, Lau Jia Jaw, Zhang Hui, Bryan Kian Hsiang Low
Reinforcement Learning with Human Feedback (RLHF) has been demonstrated to significantly enhance the performance of large language models (LLMs) by aligning their outputs with desired human values through instruction tuning.
no code implementations • 26 Jan 2023 • Flint Xiaofeng Fan, Yining Ma, Zhongxiang Dai, Cheston Tan, Bryan Kian Hsiang Low, Roger Wattenhofer
Federated Reinforcement Learning (FedRL) encourages distributed agents to learn collectively from each other's experience to improve their performance without exchanging their raw trajectories.
1 code implementation • 13 Oct 2022 • Zhongxiang Dai, Yao Shu, Bryan Kian Hsiang Low, Patrick Jaillet
linear model), which is equivalently sampled from the GP posterior with the NTK as the kernel function.
1 code implementation • 19 Jun 2022 • Arun Verma, Zhongxiang Dai, Bryan Kian Hsiang Low
The existing BO methods assume that the function evaluation (feedback) is available to the learner immediately or after a fixed delay.
1 code implementation • 14 Jun 2022 • Zhongxiang Dai, Yizhou Chen, Haibin Yu, Bryan Kian Hsiang Low, Patrick Jaillet
We prove that both algorithms are asymptotically no-regret even when some or all previous tasks are dissimilar to the current task, and show that RM-GP-UCB enjoys a better theoretical robustness than RM-GP-TS.
1 code implementation • 28 May 2022 • Zhongxiang Dai, Yao Shu, Arun Verma, Flint Xiaofeng Fan, Bryan Kian Hsiang Low, Patrick Jaillet
To better exploit the federated setting, FN-UCB adopts a weighted combination of two UCBs: $\text{UCB}^{a}$ allows every agent to additionally use the observations from the other agents to accelerate exploration (without sharing raw observations), while $\text{UCB}^{b}$ uses an NN with aggregated parameters for reward prediction in a similar way to federated averaging for supervised learning.
1 code implementation • 16 May 2022 • Lucas Agussurja, Xinyi Xu, Bryan Kian Hsiang Low
We show that for any two players, under some regularity conditions, their difference in Shapley value converges in probability to the difference in Shapley value of a limiting game whose characteristic function is proportional to the log-determinant of the joint Fisher information.
no code implementations • 10 May 2022 • Shouri Hu, Haowei Wang, Zhongxiang Dai, Bryan Kian Hsiang Low, Szu Hui Ng
To adapt the EI for better performance under cumulative regret, we introduce a novel quantity called the evaluation cost which is compared against the acquisition function, and with this, develop the expected improvement-cost (EIC) algorithm.
no code implementations • 28 Feb 2022 • Quoc Phong Nguyen, Ryutaro Oikawa, Dinil Mon Divakaran, Mun Choon Chan, Bryan Kian Hsiang Low
Similarly, MCU can be used to erase the lineage of a user's personal data from trained ML models, thus upholding a user's "right to be forgotten".
no code implementations • 28 Feb 2022 • Quoc Phong Nguyen, Bryan Kian Hsiang Low, Patrick Jaillet
Although the existing max-value entropy search (MES) is based on the widely celebrated notion of mutual information, its empirical performance can suffer due to two misconceptions whose implications on the exploration-exploitation trade-off are investigated in this paper.
1 code implementation • 24 Jan 2022 • Yao Shu, Zhongxiang Dai, Zhaoxuan Wu, Bryan Kian Hsiang Low
As a consequence, (a) the relationships among these metrics are unclear, (b) there is no theoretical interpretation for their empirical performances, and (c) there may exist untapped potential in existing training-free NAS, which probably can be unveiled through a unified theoretical understanding.
1 code implementation • 17 Dec 2021 • Sebastian Shenghong Tay, Xinyi Xu, Chuan Sheng Foo, Bryan Kian Hsiang Low
This paper presents a novel collaborative generative modeling (CGM) framework that incentivizes collaboration among self-interested parties to contribute data to a pool for training a generative model (e. g., GAN), from which synthetic data are drawn and distributed to the parties as rewards commensurate to their contributions.
1 code implementation • NeurIPS 2021 • Quoc Phong Nguyen, Zhongxiang Dai, Bryan Kian Hsiang Low, Patrick Jaillet
This paper presents two Bayesian optimization (BO) algorithms with theoretical performance guarantee to maximize the conditional value-at-risk (CVaR) of a black-box function: CV-UCB and CV-TS which are based on the well-established principle of optimism in the face of uncertainty and Thompson sampling, respectively.
no code implementations • NeurIPS 2021 • Xinyi Xu, Lingjuan Lyu, Xingjun Ma, Chenglin Miao, Chuan Sheng Foo, Bryan Kian Hsiang Low
In this paper, we adopt federated learning as a gradient-based formalization of collaborative machine learning, propose a novel cosine gradient Shapley value to evaluate the agents’ uploaded model parameter updates/gradients, and design theoretically guaranteed fair rewards in the form of better model performance.
no code implementations • NeurIPS 2021 • Xinyi Xu, Zhaoxuan Wu, Chuan Sheng Foo, Bryan Kian Hsiang Low
We observe that the diversity of the data points is an inherent property of the dataset that is independent of validation.
no code implementations • NeurIPS 2021 • Zhongxiang Dai, Bryan Kian Hsiang Low, Patrick Jaillet
The resulting differentially private FTS with DE (DP-FTS-DE) algorithm is endowed with theoretical guarantees for both the privacy and utility and is amenable to interesting theoretical insights about the privacy-utility trade-off.
2 code implementations • NeurIPS 2021 • Flint Xiaofeng Fan, Yining Ma, Zhongxiang Dai, Wei Jing, Cheston Tan, Bryan Kian Hsiang Low
The growing literature of Federated Learning (FL) has recently inspired Federated Reinforcement Learning (FRL) to encourage multiple agents to federatively build a better decision-making policy without sharing raw trajectories.
no code implementations • 6 Sep 2021 • Yao Shu, Yizhou Chen, Zhongxiang Dai, Bryan Kian Hsiang Low
Unfortunately, these NAS algorithms aim to select only one single well-performing architecture from their search spaces and thus have overlooked the capability of neural network ensemble (i. e., an ensemble of neural networks with diverse architectures) in achieving improved performance over a single final selected architecture.
no code implementations • ICLR 2022 • Yao Shu, Shaofeng Cai, Zhongxiang Dai, Beng Chin Ooi, Bryan Kian Hsiang Low
Recent years have witnessed a surging interest in Neural Architecture Search (NAS).
1 code implementation • 30 Jul 2021 • Quoc Phong Nguyen, Zhaoxuan Wu, Bryan Kian Hsiang Low, Patrick Jaillet
Information-based Bayesian optimization (BO) algorithms have achieved state-of-the-art performance in optimizing a black-box objective function.
no code implementations • 13 May 2021 • Quoc Phong Nguyen, Zhongxiang Dai, Bryan Kian Hsiang Low, Patrick Jaillet
Value-at-risk (VaR) is an established measure to assess risks in critical real-world applications with random environmental factors.
no code implementations • 17 Apr 2021 • Haibin Yu, Dapeng Liu, Yizhou Chen, Bryan Kian Hsiang Low, Patrick Jaillet
Deep Gaussian processes (DGPs), a hierarchical composition of GP models, have successfully boosted the expressive power of their single-layer counterpart.
no code implementations • 1 Jan 2021 • Yizhou Chen, Dong Li, Na Li, TONG LIANG, Shizhuo Zhang, Bryan Kian Hsiang Low
This paper presents a novel implicit process-based meta-learning (IPML) algorithm that, in contrast to existing works, explicitly represents each task as a continuous latent vector and models its probabilistic belief within the highly expressive IP framework.
no code implementations • 1 Jan 2021 • Mohit Rajpal, Yehong Zhang, Bryan Kian Hsiang Low
Pruning is an approach to alleviate overparameterization of deep neural networks (DNN) by zeroing out or pruning DNN elements with little to no efficacy at a given task.
1 code implementation • 19 Dec 2020 • Quoc Phong Nguyen, Sebastian Tay, Bryan Kian Hsiang Low, Patrick Jaillet
This paper presents a novel approach to top-$k$ ranking Bayesian optimization (top-$k$ ranking BO) which is a practical and significant generalization of preferential BO to handle top-$k$ ranking and tie/indifference observations.
1 code implementation • 19 Dec 2020 • Quoc Phong Nguyen, Bryan Kian Hsiang Low, Patrick Jaillet
This paper presents an information-theoretic framework for unifying active learning problems: level set estimation (LSE), Bayesian optimization (BO), and their generalized variant.
no code implementations • NeurIPS 2020 • Sreejith Balakrishnan, Quoc Phong Nguyen, Bryan Kian Hsiang Low, Harold Soh
The problem of inverse reinforcement learning (IRL) is relevant to a variety of tasks including value alignment and robot learning from demonstration.
no code implementations • NeurIPS 2020 • Quoc Phong Nguyen, Bryan Kian Hsiang Low, Patrick Jaillet
We frame this problem as one of minimizing the Kullback-Leibler divergence between the approximate posterior belief of model parameters after directly unlearning from erased data vs. the exact posterior belief from retraining with remaining data.
no code implementations • ICML 2020 • Rachael Hwee Ling Sim, Yehong Zhang, Mun Choon Chan, Bryan Kian Hsiang Low
This paper proposes to value a party's reward based on Shapley value and information gain on model parameters given its data.
no code implementations • 24 Oct 2020 • Dmitrii Kharkovskii, Zhongxiang Dai, Bryan Kian Hsiang Low
This paper presents the private-outsourced-Gaussian process-upper confidence bound (PO-GP-UCB) algorithm, which is the first algorithm for privacy-preserving Bayesian optimization (BO) in the outsourced setting with a provable performance guarantee.
no code implementations • 23 Oct 2019 • Mohit Rajpal, Bryan Kian Hsiang Low
This paper presents a novel unifying framework of bilinear LSTMs that can represent and utilize the nonlinear interaction of the input features present in sequence datasets for achieving superior performance over a linear LSTM and yet not incur more parameters to be learned.
no code implementations • ICML 2017 • Erik A. Daxberger, Bryan Kian Hsiang Low
To realize this, we generalize GP-UCB to a new batch variant amenable to a Markov approximation, which can then be naturally formulated as a multi-agent distributed constraint optimization problem in order to fully exploit the efficiency of its state-of-the-art solvers for achieving linear time in the batch size.
no code implementations • NeurIPS 2015 • Quoc Phong Nguyen, Bryan Kian Hsiang Low, Patrick Jaillet
By representing our IRL problem with a probabilistic graphical model, an expectation-maximization (EM) algorithm can be devised to iteratively learn the different reward functions and the stochastic transitions between them in order to jointly improve the likelihood of the expert’s demonstrated trajectories.