Clean-image Backdoor Attacks

To gather a significant quantity of annotated training data for high-performance image classification models, numerous companies opt to enlist third-party providers to label their unlabeled data. This practice is widely regarded as secure, even in cases where some annotated errors occur, as the impact of these minor inaccuracies on the final performance of the models is negligible and existing backdoor attacks require attacker's ability to poison the training images. Nevertheless, in this paper, we propose clean-image backdoor attacks which uncover that backdoors can still be injected via a fraction of incorrect labels without modifying the training images. Specifically, in our attacks, the attacker first seeks a trigger feature to divide the training images into two parts: those with the feature and those without it. Subsequently, the attacker falsifies the labels of the former part to a backdoor class. The backdoor will be finally implanted into the target model after it is trained on the poisoned data. During the inference phase, the attacker can activate the backdoor in two ways: slightly modifying the input image to obtain the trigger feature, or taking an image that naturally has the trigger feature as input. We conduct extensive experiments to demonstrate the effectiveness and practicality of our attacks. According to the experimental results, we conclude that our attacks seriously jeopardize the fairness and robustness of image classification models, and it is necessary to be vigilant about the incorrect labels in outsourced labeling.

CoMeta: Enhancing Meta Embeddings with Collaborative Information in Cold-start Problem of Recommendation

The cold-start problem is quite challenging for existing recommendation models. Specifically, for the new items with only a few interactions, their ID embeddings are trained inadequately, leading to poor recommendation performance. Some recent studies introduce meta learning to solve the cold-start problem by generating meta embeddings for new items as their initial ID embeddings. However, we argue that the capability of these methods is limited, because they mainly utilize item attribute features which only contain little information, but ignore the useful collaborative information contained in the ID embeddings of users and old items. To tackle this issue, we propose CoMeta to enhance the meta embeddings with the collaborative information. CoMeta consists of two submodules: B-EG and S-EG. Specifically, for a new item: B-EG calculates the similarity-based weighted sum of the ID embeddings of old items as its base embedding; S-EG generates its shift embedding not only with its attribute features but also with the average ID embedding of the users who interacted with it. The final meta embedding is obtained by adding up the base embedding and the shift embedding. We conduct extensive experiments on two public datasets. The experimental results demonstrate both the effectiveness and the compatibility of CoMeta.

Poisoning Deep Learning based Recommender Model in Federated Learning Scenarios

Various attack methods against recommender systems have been proposed in the past years, and the security issues of recommender systems have drawn considerable attention. Traditional attacks attempt to make target items recommended to as many users as possible by poisoning the training data. Benifiting from the feature of protecting users' private data, federated recommendation can effectively defend such attacks. Therefore, quite a few works have devoted themselves to developing federated recommender systems. For proving current federated recommendation is still vulnerable, in this work we probe to design attack approaches targeting deep learning based recommender models in federated learning scenarios. Specifically, our attacks generate poisoned gradients for manipulated malicious users to upload based on two strategies (i.e., random approximation and hard user mining). Extensive experiments show that our well-designed attacks can effectively poison the target models, and the attack effectiveness sets the state-of-the-art.

FedRecAttack: Model Poisoning Attack to Federated Recommendation

Federated Recommendation (FR) has received considerable popularity and attention in the past few years. In FR, for each user, its feature vector and interaction data are kept locally on its own client thus are private to others. Without the access to above information, most existing poisoning attacks against recommender systems or federated learning lose validity. Benifiting from this characteristic, FR is commonly considered fairly secured. However, we argue that there is still possible and necessary security improvement could be made in FR. To prove our opinion, in this paper we present FedRecAttack, a model poisoning attack to FR aiming to raise the exposure ratio of target items. In most recommendation scenarios, apart from private user-item interactions (e.g., clicks, watches and purchases), some interactions are public (e.g., likes, follows and comments). Motivated by this point, in FedRecAttack we make use of the public interactions to approximate users' feature vectors, thereby attacker can generate poisoned gradients accordingly and control malicious users to upload the poisoned gradients in a well-designed way. To evaluate the effectiveness and side effects of FedRecAttack, we conduct extensive experiments on three real-world datasets of different sizes from two completely different scenarios. Experimental results demonstrate that our proposed FedRecAttack achieves the state-of-the-art effectiveness while its side effects are negligible. Moreover, even with small proportion (3%) of malicious users and small proportion (1%) of public interactions, FedRecAttack remains highly effective, which reveals that FR is more vulnerable to attack than people commonly considered.