arXiv subjects

Le Feng

Publications and source records attributed to Le Feng.

3 recordsLinked to original sources

arXiv2025

WSD-MIL: Window Scale Decay Multiple Instance Learning for Whole Slide Image Classification

In recent years, the integration of pre-trained foundational models with multiple instance learning (MIL) has improved diagnostic accuracy in computational pathology. However, existing MIL methods focus on optimizing feature extractors and aggregation strategies while overlooking the complex semantic relationships among instances within whole slide image (WSI). Although Transformer-based MIL approaches aiming to model instance dependencies, the quadratic computational complexity limits their scalability to large-scale WSIs. Moreover, due to the pronounced variations in tumor region scales across different WSIs, existing Transformer-based methods employing fixed-scale attention mechanisms face significant challenges in precisely capturing local instance correlations and fail to account for the distance-based decay effect of patch relevance. To address these challenges, we propose window scale decay MIL (WSD-MIL), designed to enhance the capacity to model tumor regions of varying scales while improving computational efficiency. WSD-MIL comprises: 1) a window scale decay based attention module, which employs a cluster-based sampling strategy to reduce computational costs while progressively decaying attention window-scale to capture local instance relationships at varying scales; and 2) a squeeze-and-excitation based region gate module, which dynamically adjusts window weights to enhance global information modeling. Experimental results demonstrate that WSD-MIL achieves state-of-the-art performance on the CAMELYON16 and TCGA-BRCA datasets while reducing 62% of the computational memory. The code will be publicly available.

cs.CV↗

arXiv2023

Trojaning semi-supervised learning model via poisoning wild images on the web

Wild images on the web are vulnerable to backdoor (also called trojan) poisoning, causing machine learning models learned on these images to be injected with backdoors. Most previous attacks assumed that the wild images are labeled. In reality, however, most images on the web are unlabeled. Specifically, we study the effects of unlabeled backdoor images under semi-supervised learning (SSL) on widely studied deep neural networks. To be realistic, we assume that the adversary is zero-knowledge and that the semi-supervised learning model is trained from scratch. Firstly, we find the fact that backdoor poisoning always fails when poisoned unlabeled images come from different classes, which is different from poisoning the labeled images. The reason is that the SSL algorithms always strive to correct them during training. Therefore, for unlabeled images, we implement backdoor poisoning on images from the target class. Then, we propose a gradient matching strategy to craft poisoned images such that their gradients match the gradients of target images on the SSL model, which can fit poisoned images to the target class and realize backdoor injection. To the best of our knowledge, this may be the first approach to backdoor poisoning on unlabeled images of trained-from-scratch SSL models. Experiments show that our poisoning achieves state-of-the-art attack success rates on most SSL algorithms while bypassing modern backdoor defenses.

cs.CY↗

arXiv2021

Phase transitions in 3D Ising model with cluster weight by Monte Carlo method

A cluster weight Ising model is proposed by introducing an additional cluster weight in the partition function of the traditional Ising model. It is equivalent to the O($n$) loop model or $n$-component face cubic loop model on the two-dimensional lattice, but on the three-dimensional lattice, it is still not very clear whether or not these models have the same universality. In order to simulate the cluster weight Ising model and search for new universality class, we apply a cluster algorithm, by combining the color-assignation and the Swendsen-Wang methods. The dynamical exponent for the absolute magnetization is estimated to be $z=0.45(3)$ at $n=1.5$, consistent with that of the traditional Swendsen-Wang methods. The numerical estimation of the thermal exponent $y_t$ and magnetic exponent $y_m$, show that the universalities of the two models on the three-dimensional lattice are different. We obtain the global phase diagram containing paramagnetic and ferromagnetic phases. The phase transition between the two phases are second order at $1\leq n< n_c$ and first order at $n\geq n_c$, where $n_c\approx 2$. The scaling dimension $y_t$ equals to the system dimension $d$ when the first-order transition occurs. Our results are helpful in the understanding of some traditional statistical mechanics models.

cond-mat.stat-mech↗