Deep learning has witnessed significant advancements in recent years at the cost of increasing training, inference, and model storage overhead. While existing model compression methods strive to reduce the number of model parameters while maintaining high accuracy, they inevitably necessitate the re-training of the compressed model or impose architectural constraints. To overcome these limitations, this paper presents a novel framework, termed \textbf{K}nowledge \textbf{T}ranslation (KT), wherein a ``translation'' model is trained to receive the parameters of a larger model and generate compressed parameters. The concept of KT draws inspiration from language translation, which effectively employs neural networks to convert different languages, maintaining identical meaning. Accordingly, we explore the potential of neural networks to convert models of disparate sizes, while preserving their functionality. We propose a comprehensive framework for KT, introduce data augmentation strategies to enhance model performance despite restricted training data, and successfully demonstrate the feasibility of KT on the MNIST dataset. Code is available at \url{https://github.com/zju-SWJ/KT}.
Sampling from diffusion models can be treated as solving the corresponding ordinary differential equations (ODEs), with the aim of obtaining an accurate solution with as few number of function evaluations (NFE) as possible. Recently, various fast samplers utilizing higher-order ODE solvers have emerged and achieved better performance than the initial first-order one. However, these numerical methods inherently result in certain approximation errors, which significantly degrades sample quality with extremely small NFE (e.g., around 5). In contrast, based on the geometric observation that each sampling trajectory almost lies in a two-dimensional subspace embedded in the ambient space, we propose Approximate MEan-Direction Solver (AMED-Solver) that eliminates truncation errors by directly learning the mean direction for fast diffusion sampling. Besides, our method can be easily used as a plugin to further improve existing ODE-based samplers. Extensive experiments on image synthesis with the resolution ranging from 32 to 256 demonstrate the effectiveness of our method. With only 5 NFE, we achieve 7.14 FID on CIFAR-10, 13.75 FID on ImageNet 64$\times$64, and 12.79 FID on LSUN Bedroom. Our code is available at https://github.com/zhyzhouu/amed-solver.
Data-free knowledge distillation (DFKD) aims to obtain a lightweight student model without original training data. Existing works generally synthesize data from the pre-trained teacher model to replace the original training data for student learning. To more effectively train the student model, the synthetic data shall be customized to the current student learning ability. However, this is ignored in the existing DFKD methods and thus negatively affects the student training. To address this issue, we propose Customizing Synthetic Data for Data-Free Student Learning (CSD) in this paper, which achieves adaptive data synthesis using a self-supervised augmented auxiliary task to estimate the student learning ability. Specifically, data synthesis is dynamically adjusted to enlarge the cross entropy between the labels and the predictions from the self-supervised augmented task, thus generating hard samples for the student model. The experiments on various datasets and teacher-student models show the effectiveness of our proposed method. Code is available at: $\href{https://github.com/luoshiya/CSD}{https://github.com/luoshiya/CSD}$
Multi-Teacher knowledge distillation provides students with additional supervision from multiple pre-trained teachers with diverse information sources. Most existing methods explore different weighting strategies to obtain a powerful ensemble teacher, while ignoring the student with poor learning ability may not benefit from such specialized integrated knowledge. To address this problem, we propose Adaptive Multi-teacher Knowledge Distillation with Meta-Learning (MMKD) to supervise student with appropriate knowledge from a tailored ensemble teacher. With the help of a meta-weight network, the diverse yet compatible teacher knowledge in the output layer and intermediate layers is jointly leveraged to enhance the student performance. Extensive experiments on multiple benchmark datasets validate the effectiveness and flexibility of our methods. Code is available: https://github.com/Rorozhl/MMKD.
Recent years have witnessed significant progress in developing efficient training and fast sampling approaches for diffusion models. A recent remarkable advancement is the use of stochastic differential equations (SDEs) to describe data perturbation and generative modeling in a unified mathematical framework. In this paper, we reveal several intriguing geometric structures of diffusion models and contribute a simple yet powerful interpretation to their sampling dynamics. Through carefully inspecting a popular variance-exploding SDE and its marginal-preserving ordinary differential equation (ODE) for sampling, we discover that the data distribution and the noise distribution are smoothly connected with an explicit, quasi-linear sampling trajectory, and another implicit denoising trajectory, which even converges faster in terms of visual quality. We also establish a theoretical relationship between the optimal ODE-based sampling and the classic mean-shift (mode-seeking) algorithm, with which we can characterize the asymptotic behavior of diffusion models and identify the score deviation. These new geometric observations enable us to improve previous sampling algorithms, re-examine latent interpolation, as well as re-explain the working principles of distillation-based fast sampling techniques.
Although diffusion model has shown great potential for generating higher quality images than GANs, slow sampling speed hinders its wide application in practice. Progressive distillation is thus proposed for fast sampling by progressively aligning output images of $N$-step teacher sampler with $N/2$-step student sampler. In this paper, we argue that this distillation-based accelerating method can be further improved, especially for few-step samplers, with our proposed \textbf{C}lassifier-based \textbf{F}eature \textbf{D}istillation (CFD). Instead of aligning output images, we distill teacher's sharpened feature distribution into the student with a dataset-independent classifier, making the student focus on those important features to improve performance. We also introduce a dataset-oriented loss to further optimize the model. Experiments on CIFAR-10 show the superiority of our method in achieving high quality and fast sampling. Code will be released soon.
Graph neural networks (GNNs) have become one of the most popular research topics in both academia and industry communities for their strong ability in handling irregular graph data. However, large-scale datasets are posing great challenges for deploying GNNs in edge devices with limited resources and model compression techniques have drawn considerable research attention. Existing model compression techniques such as knowledge distillation (KD) mainly focus on convolutional neural networks (CNNs). Only limited attempts have been made recently for distilling knowledge from GNNs in an offline manner. As the performance of the teacher model does not necessarily improve as the number of layers increases in GNNs, selecting an appropriate teacher model will require substantial efforts. To address these challenges, we propose a novel online knowledge distillation framework called Alignahead++ in this paper. Alignahead++ transfers structure and feature information in a student layer to the previous layer of another simultaneously trained student model in an alternating training procedure. Meanwhile, to avoid over-smoothing problem in GNNs, deep supervision is employed in Alignahead++ by adding an auxiliary classifier in each intermediate layer to prevent the collapse of the node feature embeddings. Experimental results on four datasets including PPI, Cora, PubMed and CiteSeer demonstrate that the student performance is consistently boosted in our collaborative training framework without the supervision of a pre-trained teacher model and its effectiveness can generally be improved by increasing the number of students.
Considerable progress has been made in domain generalization (DG) which aims to learn a generalizable model from multiple well-annotated source domains to unknown target domains. However, it can be prohibitively expensive to obtain sufficient annotation for source datasets in many real scenarios. To escape from the dilemma between domain generalization and annotation costs, in this paper, we introduce a novel task named label-efficient domain generalization (LEDG) to enable model generalization with label-limited source domains. To address this challenging task, we propose a novel framework called Collaborative Exploration and Generalization (CEG) which jointly optimizes active exploration and semi-supervised generalization. Specifically, in active exploration, to explore class and domain discriminability while avoiding information divergence and redundancy, we query the labels of the samples with the highest overall ranking of class uncertainty, domain representativeness, and information diversity. In semi-supervised generalization, we design MixUp-based intra- and inter-domain knowledge augmentation to expand domain knowledge and generalize domain invariance. We unify active exploration and semi-supervised generalization in a collaborative way and promote mutual enhancement between them, boosting model generalization with limited annotation. Extensive experiments show that CEG yields superior generalization performance. In particular, CEG can even use only 5% data annotation budget to achieve competitive results compared to the previous DG methods with fully labeled data on PACS dataset.
Knowledge graph embedding (KGE) has been intensively investigated for link prediction by projecting entities and relations into continuous vector spaces. Current popular high-dimensional KGE methods obtain quite slight performance gains while require enormous computation and memory costs. In contrast to high-dimensional KGE models, training low-dimensional models is more efficient and worthwhile for better deployments to practical intelligent systems. However, the model expressiveness of semantic information in knowledge graphs (KGs) is highly limited in the low dimension parameter space. In this paper, we propose iterative self-semantic knowledge distillation strategy to improve the KGE model expressiveness in the low dimension space. KGE model combined with our proposed strategy plays the teacher and student roles alternatively during the whole training process. Specifically, at a certain iteration, the model is regarded as a teacher to provide semantic information for the student. At next iteration, the model is regard as a student to incorporate the semantic information transferred from the teacher. We also design a novel semantic extraction block to extract iteration-based semantic information for the training model self-distillation. Iteratively incorporating and accumulating iteration-based semantic information enables the low-dimensional model to be more expressive for better link prediction in KGs. There is only one model during the whole training, which alleviates the increase of computational expensiveness and memory requirements. Furthermore, the proposed strategy is model-agnostic and can be seamlessly combined with other KGE models. Consistent and significant performance gains in experimental evaluations on four standard datasets demonstrate the effectiveness of the proposed self-distillation strategy.
Existing knowledge distillation methods on graph neural networks (GNNs) are almost offline, where the student model extracts knowledge from a powerful teacher model to improve its performance. However, a pre-trained teacher model is not always accessible due to training cost, privacy, etc. In this paper, we propose a novel online knowledge distillation framework to resolve this problem. Specifically, each student GNN model learns the extracted local structure from another simultaneously trained counterpart in an alternating training procedure. We further develop a cross-layer distillation strategy by aligning ahead one student layer with the layer in different depth of another student model, which theoretically makes the structure information spread over all layers. Experimental results on five datasets including PPI, Coauthor-CS/Physics and Amazon-Computer/Photo demonstrate that the student performance is consistently boosted in our collaborative training framework without the supervision of a pre-trained teacher model. In addition, we also find that our alignahead technique can accelerate the model convergence speed and its effectiveness can be generally improved by increasing the student numbers in training. Code is available: https://github.com/GuoJY-eatsTG/Alignahead