In offline reinforcement learning, it is necessary to manage out-of-distribution actions to prevent overestimation of value functions. One class of methods, the policy-regularized method, addresses this problem by constraining the target policy to stay close to the behavior policy. Although several approaches suggest representing the behavior policy as an expressive diffusion model to boost performance, it remains unclear how to regularize the target policy given a diffusion-modeled behavior sampler. In this paper, we propose Diffusion Actor-Critic (DAC) that formulates the Kullback-Leibler (KL) constraint policy iteration as a diffusion noise regression problem, enabling direct representation of target policies as diffusion models. Our approach follows the actor-critic learning paradigm in which we alternatively train a diffusion-modeled target policy and a critic network. The actor training loss includes a soft Q-guidance term from the Q-gradient. The soft Q-guidance is based on the theoretical solution of the KL constraint policy iteration, which prevents the learned policy from taking out-of-distribution actions. We demonstrate that such diffusion-based policy constraint, along with the coupling of the lower confidence bound of the Q-ensemble as value targets, not only preserves the multi-modality of target policies, but also contributes to stable convergence and strong performance in DAC. Our approach is evaluated on D4RL benchmarks and outperforms the state-of-the-art in nearly all environments.

Knowledge distillation (KD) is a promising yet challenging model compression approach that transmits rich learning representations from robust but resource-demanding teacher models to efficient student models. Previous methods for image super-resolution (SR) are often tailored to specific teacher-student architectures, limiting their potential for improvement and hindering broader applications. This work presents a novel KD framework for SR models, the multi-granularity Mixture of Priors Knowledge Distillation (MiPKD), which can be universally applied to a wide range of architectures at both feature and block levels. The teacher’s knowledge is effectively integrated with the student's feature via the Feature Prior Mixer, and the reconstructed feature propagates dynamically in the training phase with the Block Prior Mixer. Extensive experiments illustrate the significance of the proposed MiPKD technique.

Knowledge distillation (KD) compresses deep neural networks by transferring task-related knowledge from cumbersome pre-trained teacher models to more compact student models. However, vanilla KD for image super-resolution (SR) networks yields only limited improvements due to the inherent nature of SR tasks, where the outputs of teacher models are noisy approximations of high-quality label images. In this work, we show that the potential of vanilla KD has been underestimated and demonstrate that the ingenious application of data augmentation methods can close the gap between it and more complex, well-designed methods. Unlike conventional training processes typically applying image augmentations simultaneously to both low-quality inputs and high-quality labels, we propose AugKD utilizing unpaired data augmentations to 1) generate auxiliary distillation samples and 2) impose label consistency regularization. Comprehensive experiments show that the AugKD significantly outperforms existing state-of-the-art KD methods across a range of SR tasks.

The dramatically increasing volume of incomplete data makes the imputation models computationally infeasible in many real-life applications. In this paper, we propose an effective scalable imputation system named SCIS to significantly speed up the training of the differentiable generative adversarial imputation models under accuracy-guarantees for large-scale incomplete data. SCIS consists of two modules, differentiable imputation modeling (DIM) and sample size estimation (SSE). DIM leverages a new masking Sinkhorn divergence function to make an arbitrary generative adversarial imputation model differentiable, while for such a differentiable imputation model, SSE can estimate an appropriate sample size to ensure the user-specified imputation accuracy of the final model. Moreover, SCIS can also accelerate the autoencoder based imputation models. Extensive experiments upon several real-life large-scale datasets demonstrate that, our proposed system can accelerate the generative adversarial model training by 6.23x. Using around 1.27% samples, SCIS yields competitive accuracy with the state-of-the-art imputation methods in much shorter computation time.

Tabular data is widely utilized in a wide range of real-world applications. The challenge of few-shot learning with tabular data stands as a crucial problem in both industry and academia, due to the high cost or even impossibility of annotating additional samples. However, the inherent heterogeneity of tabular features, combined with the scarcity of labeled data, presents a significant challenge in tabular few-shot classification. In this paper, we propose a novel approach named Diffusion-based Representation with Random Distance matching (D2R2) for tabular few-shot learning. D2R2 leverages the powerful expression ability of diffusion models to extract essential semantic knowledge crucial for denoising process. This semantic knowledge proves beneficial in few-shot downstream tasks. During the training process of our designed diffusion model, we introduce a random distance matching to preserve distance information in the embeddings, thereby improving effectiveness for classification. During the classification stage, we introduce an instance-wise iterative prototype scheme to improve performance by accommodating the multimodality of embeddings and increasing clustering robustness. Our experiments reveal the significant efficacy of D2R2 across various tabular few-shot learning benchmarks, demonstrating its state-of-the-art performance in this field.

``Distribution shift'' is the primary obstacle to the success of offline reinforcement learning. As a learning policy may take actions beyond the knowledge of the behavior policy (referred to as Out-of-Distribution (OOD) actions), the Q-values of these OOD actions can be easily overestimated. Consequently, the learning policy becomes biasedly optimized using the incorrect recovered Q-value function. One commonly used idea to avoid the overestimation of Q-value is to make a pessimistic adjustment. Our key idea is to penalize the Q-values of OOD actions that correspond to high uncertainty. In this work, we propose Q-Distribution guided Q-learning (QDQ) which pessimistic Q-value on OOD regions based on uncertainty estimation. The uncertainty measure is based on the conditional Q-value distribution, which is learned via a high-fidelity and efficient consistency model. On the other hand, to avoid the overly conservative problem, we introduce an uncertainty-aware optimization objective to update the Q-value function. The proposed QDQ demonstrates solid theoretical guarantees for the accuracy of Q-value distribution learning and uncertainty measurement, as well as the performance of the learning policy. QDQ consistently exhibits strong performance in the D4RL benchmark and shows significant improvements for many tasks. Our code can be found at <code link>.

MOTIVATION: Effective drug delivery systems are paramount in enhancing pharmaceutical outcomes, particularly through the use of cell-penetrating peptides (CPPs). These peptides are gaining prominence due to their ability to penetrate eukaryotic cells efficiently without inflicting significant damage to the cellular membrane, thereby ensuring optimal drug delivery. However, the identification and characterization of CPPs remain a challenge due to the laborious and time-consuming nature of conventional methods, despite advances in proteomics. Current computational models, however, are predominantly tailored for balanced datasets, an approach that falls short in real-world applications characterized by a scarcity of known positive CPP instances. RESULTS: To navigate this shortfall, we introduce PractiCPP, a novel deep-learning framework tailored for CPP prediction in highly imbalanced data scenarios. Uniquely designed with the integration of hard negative sampling and a sophisticated feature extraction and prediction module, PractiCPP facilitates an intricate understanding and learning from imbalanced data. Our extensive computational validations highlight PractiCPP's exceptional ability to outperform existing state-of-the-art methods, demonstrating remarkable accuracy, even in datasets with an extreme positive-to-negative ratio of 1:1000. Furthermore, through methodical embedding visualizations, we have established that models trained on balanced datasets are not conducive to practical, large-scale CPP identification, as they do not accurately reflect real-world complexities. In summary, PractiCPP potentially offers new perspectives in CPP prediction methodologies. Its design and validation, informed by real-world dataset constraints, suggest its utility as a valuable tool in supporting the acceleration of drug delivery advancements. AVAILABILITY AND IMPLEMENTATION: The source code of PractiCPP is available on Figshare at https://doi.org/10.6084/m9.figshare.25053878.v1.

Data imputation has been extensively explored to solve the missing data problem. The dramatically increasing volume of incomplete data makes the imputation models computationally infeasible in many real-life applications. In this paper, we propose an effective scalable imputation system named SCIS to significantly speed up the training of the differentiable generative adversarial imputation models under accuracy-guarantees for large-scale incomplete data. SCIS consists of two modules,differentiable imputation modeling(DIM) andsample size estimation(SSE). DIM leverages a newmasking Sinkhorndivergence function to make an arbitrary generative adversarial imputation model differentiable, while for such a differentiable imputation model, SSE can estimate an appropriate sample size to ensure the user-specified imputation accuracy of the final model. Moreover, SCIS can also accelerate the autoencoder based imputation models. Extensive experiments upon several real-life large-scale datasets demonstrate that, our proposed system can accelerate the generative adversarial model training by 6.23x. Using around 1.27% samples, SCIS yields competitive accuracy with the state-of-the-art imputation methods in much shorter computation time.

Due to the inability to interact with the environment, offline reinforcement learning (RL) methods face the challenge of estimating the Out-of-Distribution (OOD) points. Existing methods for addressing this issue either control policy to exclude the OOD action or make the $Q$ function pessimistic. However, these methods can be overly conservative or fail to identify OOD areas accurately. To overcome this problem, we propose a Constrained Policy optimization with Explicit Behavior density (CPED) method that utilizes a flow-GAN model to explicitly estimate the density of behavior policy. By estimating the explicit density, CPED can accurately identify the safe region and enable exploration within the region, resulting in less conservative learning policies. We further provide theoretical results for both the flow-GAN estimator and performance guarantee for CPED by showing that CPED can find the optimal $Q$-function value. Empirically, CPED outperforms existing alternatives on various standard offline reinforcement learning tasks, yielding higher expected returns.

Gaussian process regression is widely used in many fields, for example, machine learning, reinforcement learning and uncertainty quantification. One key component of Gaussian process regression is the unknown correlation function, which needs to be specified. In this paper, we investigate what would happen if the correlation function is misspecified. We derive upper and lower error bounds for Gaussian process regression with possibly misspecified correlation functions. We find that when the sampling scheme is quasi-uniform, the optimal convergence rate can be attained even if the smoothness of the imposed correlation function exceeds that of the true correlation function. We also obtain convergence rates of kernel ridge regression with misspecified kernel function, where the underlying truth is a deterministic function. Our study reveals a close connection between the convergence rates of Gaussian process regression and kernel ridge regression, which is aligned with the relationship between sample paths of Gaussian process and the corresponding reproducing kernel Hilbert space. This work establishes a bridge between Bayesian learning based on Gaussian process and frequentist kernel methods with reproducing kernel Hilbert space.