Vaccination with inactivated (killed) whole-virus particles has been used to prevent a wide range of viral diseases. However, for an HIV vaccine this approach has been largely negated due to inherent safety concerns, despite the ability of killed whole-virus vaccines to generate a strong, predominantly antibody-mediated immune response in vivo. HIV-1 Clade B NL4-3 was genetically modified by deleting the nef and vpu genes and substituting the coding sequence for the Env signal peptide with that of honeybee melittin signal peptide to produce a less virulent and more replication efficient virus. This genetically modified virus (gmHIV-1NL4-3) was inactivated and formulated as a killed whole-HIV vaccine, and then used for a Phase I human clinical trial (Trial Registration: Clinical Trials NCT01546818). The gmHIV-1NL4-3 was propagated in the A3.01 human T cell line followed by virus purification and inactivation with aldrithiol-2 and γ-irradiation. Thirty-three HIV-1 positive volunteers receiving cART were recruited for this observer-blinded, placebo-controlled Phase I human clinical trial to assess the safety and immunogenicity. Genetically modified and killed whole-HIV-1 vaccine, SAV001, was well tolerated with no serious adverse events. HIV-1NL4-3-specific PCR showed neither evidence of vaccine virus replication in the vaccine virus-infected human T lymphocytes in vitro nor in the participating volunteers receiving SAV001 vaccine. Furthermore, SAV001 with adjuvant significantly increased the pre-existing antibody response to HIV-1 proteins. Antibodies in the plasma of vaccinees were also found to recognize HIV-1 envelope protein on the surface of infected cells as well as showing an enhancement of broadly neutralizing antibodies inhibiting tier I and II of HIV-1 B, D, and A subtypes. The killed whole-HIV vaccine, SAV001, is safe and triggers anti-HIV immune responses. It remains to be determined through an appropriate trial whether this immune response prevents HIV infection.

Convolutional neural networks (CNNs) have achieved tremendous success in solving complex classification problems. Motivated by this success, there have been proposed various compression methods for downsizing the CNNs to deploy them on resource-constrained embedded systems. However, a new type of vulnerability of compressed CNNs known as the adversarial examples has been discovered recently, which is critical for security-sensitive systems because the adversarial examples can cause malfunction of CNNs and can be crafted easily in many cases. In this paper, we proposed a compression framework to produce compressed CNNs robust against such adversarial examples. To achieve the goal, our framework uses both pruning and knowledge distillation with adversarial training. We formulate our framework as an optimization problem and provide a solution algorithm based on the proximal gradient method, which is more memory-efficient than the popular ADMM-based compression approaches. In experiments, we show that our framework can improve the trade-off between adversarial robustness and compression rate compared to the existing state-of-the-art adversarial pruning approach.

A network intrusion detection (NID) system plays a critical role in cybersecurity. However, the existing machine learning-based NID research has a vital issue that their experimental settings do not reflect real-world situations where unknown attacks are constantly emerging. In particular, their train and test sets are from a single data set, which inevitably overestimates the detection power since all test attack types are known in training, and test cases will have similar characteristics to the training data. This paper introduces a new strategy to constitute test data with updated traffic with attack types not included in training data. In the proposed setting, the prediction accuracy of the existing detectors is dropped by about 20% compared to what has been reported. Also, in- depth analysis of detection performance by attack types has revealed that the existing models have strength at certain attack types but struggle to detect the other attack types such as DoS, DDoS, web attack, and port scan. To overcome the issues, we propose a new neural detector, called MHSA, based on a multi-head self-attention mechanism whose architecture suits better to capture scattered pieces of evidence in network traffic. Our model improved the overall detection performance by 29% in false positive rate at the true positive rate of 0.9 and by 9% in AUC over the current state-of-the-art models, successfully detecting the attacks that are not well captured before. Furthermore, we show that our proposed MHSA model even outperforms the best ensemble detector constructed by joining the state-of-the-art classifiers.