With powerful parallel computing GPUs and massive user data, neural-network-based deep learning can well exert its strong power in problem modeling and solving, and has archived great success in many applications such as image classification, speech recognition and machine translation etc. While deep learning has been increasingly popular, the problem of privacy leakage becomes more and more urgent. Given the fact that the training data may contain highly sensitive information, e.g., personal medical records, directly sharing them among the users (i.e., participants) or centrally storing them in one single location may pose a considerable threat to user privacy. In this paper, we present a practical privacy-preserving collaborative deep learning system that allows users to cooperatively build a collective deep learning model with data of all participants, without direct data sharing and central data storage. In our system, each participant trains a local model with their own data and only shares model parameters with the others. To further avoid potential privacy leakage from sharing model parameters, we use functional mechanism to perturb the objective function of the neural network in the training process to achieve -differential privacy. In particular, for the first time, we consider the existence of unreliable participants, i.e., the participants with low-quality data, and propose a solution to reduce the impact of these participants while protecting their privacy. We evaluate the performance of our system on two well-known real-world datasets for regression and classification tasks. The results demonstrate that the proposed system is robust against unreliable participants, and achieves high accuracy close to the model trained in a traditional centralized manner while ensuring rigorous privacy protection.

Recent years have witnessed an exponential increase in the use of mobile and embedded devices. With the great success of deep learning in many fields, there is an emerging trend to deploy deep learning on mobile and embedded devices to better meet the requirement of real-time applications and user privacy protection. However, the limited resources of mobile and embedded devices make it challenging to fulfill the intensive computation and storage demand of deep learning models. In this survey, we conduct a comprehensive review on the related issues for deep learning on mobile and embedded devices. We start with a brief introduction of deep learning and discuss major challenges of implementing deep learning models on mobile and embedded devices. We then conduct an in-depth survey on important compression and acceleration techniques that help adapt deep learning models to mobile and embedded devices, which we specifically classify as pruning, quantization, model distillation, network design strategies, and low-rank factorization. We elaborate on the hardware-based solutions, including mobile GPU, FPGA, and ASIC, and describe software frameworks for mobile deep learning models, especially the development of frameworks based on OpenCL and RenderScript. After that, we present the application of mobile deep learning in a variety of areas, such as navigation, health, speech recognition, and information security. Finally, we discuss some future directions for deep learning on mobile and embedded devices to inspire further research in this area.

Auction is an effective way to allocate goods or services to bidders who value them the most. The rapid growth of e-auctions facilitates online transactions but poses new and distinctive challenges. It is difficult to establish trusts among sellers, buyers and auctioneers without the centralized auction websites or platforms (the auctioneer) that collect bids and derive the auction results. However, these third parties may be untrustworthy, and malicious sellers or buyers may refuse to deliver the goods or payment according to the protocol. Moreover, the open and anonymous online environment may stimulate auction participants to form collusion coalitions to rig the auction and reap unfair profit. Many auction designs have been proposed to address these concerns, but they fall short of simultaneously achieving decentralization (i.e., held without a trusted third utility), strong consensus (i.e., the establishment of trust), collusion-resistance and practical implementation. We present CReam, the first decentralized collusion-resistant e-auction system that is implemented with smart contract on the blockchain. With the carefully-designed smart auction contract, mutually distrustful and rational sellers and buyers are stimulated to operate properly hence transact safely without trusted third parties. The auction mechanism in the smart contract can effectively prevent bidder collusion and realize economic robustness, i.e., truthfulness. We implement a fully functional CReam on the Ethereum network. Extensive experimental results confirm that CReam can greatly reduce the probability of collusion and achieve an approximate optimal revenue at a low cost of contract execution.

Voice authentication is becoming increasingly popular, which offers potential benefits over knowledge and possession based authentication methods. Meanwhile, the unique features of lip movements during speaking have been proved to be useful for authentication. However, the unimodal biometric authentication systems based on either voice or lip movements have certain limitations. Voice authentication systems are prone to spoofing attacks and suffer from serious performance degradation in noisy environments. Lip movements authentication systems are unstable and are sensitive to the user’s physical and psychological conditions. In this paper, we propose and implement LVID, a multimodal biometrics authentication system on smartphones, which resolves the defects of the original systems by combining the advantages of lip movements and voice. LVID simultaneously captures these two biometrics with the built-in audio devices on smartphones and fuses them at the data level. The reliable and effective features are then extracted from the fused data for authentication. LVID is practical as it requires neither cumbersome operations nor additional hardwares but only a speaker and a microphone that are commonly available on smartphones. Our experimental results with 104 participants show that LVID can achieve 95% accuracy for user authentication, and 93.47% of the attacks can be detected. It is also verified that LVID works well with different smartphones and is robust to different smartphone positions.

Nowadays machine learning is becoming a new paradigm for mining hidden knowledge in big data. The collection and manipulation of big data not only create considerable values, but also raise serious privacy concerns. To protect the huge amount of potentially sensitive data, a straightforward approach is to encrypt data with specialized cryptographic tools. However, it is challenging to utilize or operate on encrypted data, especially to perform machine learning algorithms. In this paper, we investigate the problem of training high quality word vectors over large-scale encrypted data (from distributed data owners) with the privacy-preserving collaborative neural network learning algorithms. We leverage and also design a suite of arithmetic primitives (e.g., multiplication, fixed-point representation and sigmoid function computation etc.) on encrypted data, served as components of our construction. We theoretically analyze the security and efficiency of our proposed construction, and conduct extensive experiments on representative real-world datasets to verify its practicality and effectiveness.

The time and monetary costs of training sophisticated deep neural networks are exorbitant, which motivates resource-limited users to outsource the training process to the cloud. Concerning that an untrustworthy cloud service provider may inject backdoors to the returned model, the user can leverage state-of-the-art defense strategies to examine the model. In this paper, we aim to develop robust backdoor attacks (named RobNet) that can evade existing defense strategies from the standpoint of malicious cloud providers. The key rationale is to diversify the triggers and strengthen the model structure so that the backdoor is hard to be detected or removed. To attain this objective, we refine the trigger generation algorithm by selecting the neuron(s) with large weights and activations and then computing the triggers via gradient descent to maximize the value of the selected neuron(s). In stark contrast to existing works that fix the trigger location, we design a multi-location patching method to make the model less sensitive to mild displacement of triggers in real attacks. Furthermore, we extend the attack space by proposing multi-trigger backdoor attacks that can misclassify inputs with different triggers into the same or different target label(s). We evaluate the performance of RobNet on MNIST, GTSRB, and CIFAR-10 datasets, against four representative defense strategies Pruning, NeuralCleanse, Strip, and ABS. The comparison with two state-of-the-art baselines BadNets and Hidden Backdoors demonstrates that RobNet achieves higher attack success rate and is more resistant to potential defenses.