Knowledge distillation (KD) is a vital technique for deploying deep neural networks (DNNs) on resource-constrained devices by transferring knowledge from large teacher models to lightweight student models. While teacher models from third-party platforms may undergo security verification (e.g., backdoor detection), we uncover a novel and critical threat: distillation-conditional backdoor attacks (DCBAs). DCBA injects dormant and undetectable backdoors into teacher models, which become activated in student models via the KD process, even with clean distillation datasets. While the direct extension of existing methods is ineffective for DCBA, we implement this attack by formulating it as a bilevel optimization problem and proposing a simple yet effective method (i.e., SCAR). Specifically, the inner optimization simulates the KD process by optimizing a surrogate student model, while the outer optimization leverages outputs from this surrogate to optimize the teacher model for implanting the conditional backdoor.

We investigate the HSIC (Hilbert-Schmidt independence criterion) bottleneck as a regularizer for learning an adversarially robust deep neural network classifier. In addition to the usual cross-entropy loss, we add regularization terms for every intermediate layer to ensure that the latent representations retain useful information for output prediction while reducing redundant information. We show that the HSIC bottleneck enhances robustness to adversarial attacks both theoretically and experimentally. In particular, we prove that the HSIC bottleneck regularizer reduces the sensitivity of the classifier to adversarial examples. Our experiments on multiple benchmark datasets and architectures demonstrate that incorporating an HSIC bottleneck regularizer attains competitive natural accuracy and improves adversarial robustness, both with and without adversarial examples during training. Our code and adversarially robust models are publicly available.2

In this work, we propose a novel and surprisingly simple routing strategy called self-routing, where each capsule is routed independently by its subordinate routing network. Therefore, the agreement between capsules is not required anymore, but both poses and activations of upper-level capsules are obtained in a way similar to Mixture-of-Experts. Our experiments on CIFAR10, SVHN, and SmallNORB showthat the self-routing performs more robustly against white-box adversarial attacks and affine transformations, requiring less computation.

In this paper we consider the case of generativemodels. In particular,we investigate generativeadversarialnetworks (GANs) in the task of learning new categories in a sequential fashion.

With the disentangled representations, we synthesize the counterfactual unbiased training samples to further decorrelate causal and bias variables.

Since proxy invariance indiff approaches data.

This paper provides a unified view to explain different adversarial attacks and defensemethods,i.e.

In this paper, we present DSA, the first automated framework for discovering sparsity allocation schemes for layer-wise pruning in Large Language Models (LLMs). LLMs have become increasingly powerful, but their large parameter counts make them computationally expensive. Existing pruning methods for compressing LLMs primarily focus on evaluating redundancies and removing element-wise weights. However, these methods fail to allocate adaptive layerwise sparsities, leading to performance degradation in challenging tasks.

Graph embedding has been intensively studied recently, due to the advance of various neural networkmodels.

Program translation isanimportant tool tomigrate legacycode inone language into an ecosystem built in a different language. In this work, we are the first to employ deep neural networks toward tackling this problem.