Transfer learning provides an effective solution for feasibly and fast customize accurate Student models, by transferring the learned knowledge of pre-trained Teacher models over large datasets via fine-tuning. Many pre-trained Teacher models used in transfer learning are publicly available and maintained by public platforms, increasing their vulnerability to backdoor attacks. In this paper, we demonstrate a backdoor threat to transfer learning tasks on both image and time-series data leveraging the knowledge of publicly accessible Teacher models, aimed at defeating three commonly-adopted defenses: pruning-based, retraining-based and input pre-processing-based defenses. Specifically, (A) ranking-based selection mechanism to speed up the backdoor trigger generation and perturbation process while defeating pruning/retraining-based defenses. (B) autoencoder-powered trigger generation is proposed to produce a robust trigger that can defeat the input pre-processing-based defense. (C) defense-aware retraining to generate the manipulated model. We launch effective misclassification attacks on Student models over real-world images, brain MRI and ECG data. Experiments reveal that our attack can maintain the 98.4% and 97.2% classification accuracy as the genuine model on clean image and time series inputs respectively while improving 27.9%-100% and 27.1% -56.1% attack success rate on trojaned image and time series inputs respectively in the presence of pruning-based and/or retraining-based defenses.

Yuanchi Zhang, Yile Wang, Zijun Liu, Shuo Wang, Xiaolong Wang, Peng Li, Maosong Sun, Yang Liu. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

The futures market's forecasts are significant to investors and policymakers, where the application of deep learning approaches to finance has received a great deal of attention. In this study, we propose a multivariate financial time-series forecasting method. Our model addresses the long- and short-term features, multimodal and non-stationarity nature of multivariate time-series by incorporating the improved deep neural networks and certified noise injection. Specifically, multimodal variational autoencoder is used to extract deep high-level features of multivariate time-series, Long- and Short- Term recurrent neural network is applied for multivariate time-series forecasting, and certified noise injection mechanism, inspired by differential privacy, is proposed to improve the robustness and accuracy of prediction. Extensive empirical results on real-world agricultural commodity futures price time series and relevant external data demonstrate that our model achieves better performance over that of several state-of-the-art baseline methods.

Background: Globally the number of cancer patients and deaths are continuing to increase yearly, and cancer has, therefore, become one of the world's highest causes of morbidity and mortality. In recent years, the study of anticancer drugs has become one of the most popular medical topics. Objective: In this review, in order to study the application of machine learning in predicting anticancer drugs activity, some machine learning approaches such as Linear Discriminant Analysis (LDA), Principal components analysis (PCA), Support Vector Machine (SVM), Random forest (RF), k-Nearest Neighbor (kNN), and Naïve Bayes (NB) were selected, and the examples of their applications in anticancer drugs design are listed. Results: Machine learning contributes a lot to anticancer drugs design and helps researchers by saving time and is cost effective. However, it can only be an assisting tool for drug design. Conclusion: This paper introduces the application of machine learning approaches in anticancer drug design. Many examples of success in identification and prediction in the area of anticancer drugs activity prediction are discussed, and the anticancer drugs research is still in active progress. Moreover, the merits of some web servers related to anticancer drugs are mentioned.

Numerous open-source and commercial malware detectors are available. However, their efficacy is threatened by new adversarial attacks, whereby malware attempts to evade detection, e.g., by performing feature-space manipulation. In this work, we propose an explainability-guided and model-agnostic testing framework for robustness of malware detectors when confronted with adversarial attacks. The framework introduces the concept of Accrued Malicious Magnitude (AMM) to identify which malware features could be manipulated to maximize the likelihood of evading detection. We then use this framework to test several state-of-the-art malware detectors' ability to detect manipulated malware. We find that (i) commercial antivirus engines are vulnerable to AMM-guided test cases; (ii) the ability of a manipulated malware generated using one detector to evade detection by another detector (i.e., transferability) depends on the overlap of features with large AMM values between the different detectors; and (iii) AMM values effectively measure the fragility of features (i.e., capability of feature-space manipulation to flip the prediction results) and explain the robustness of malware detectors facing evasion attacks. Our findings shed light on the limitations of current malware detectors, as well as how they can be improved.

The susceptibility of deep neural networks (DNNs) to adversarial intrusions, exemplified by adversarial examples, is well-documented. Conventional attacks implement unstructured, pixel-wise perturbations to mislead classifiers, which often results in a noticeable departure from natural samples and lacks human-perceptible interpretability. In this work, we present an adversarial attack strategy that implements fine-granularity, semantic-meaning-oriented structural perturbations. Our proposed methodology manipulates the semantic attributes of images through the use of disentangled latent codes. We engineer adversarial perturbations by manipulating either a single latent code or a combination thereof. To this end, we propose two unsupervised semantic manipulation strategies: one based on vector-disentangled representation and the other on feature map-disentangled representation, taking into consideration the complexity of the latent codes and the smoothness of the reconstructed images. Our empirical evaluations, conducted extensively on real-world image data, showcase the potency of our attacks, particularly against black-box classifiers. Furthermore, we establish the existence of a universal semantic adversarial example that is agnostic to specific images.