Search

The integration of Differential Privacy (DP) with diffusion models (DMs) presents a promising yet challenging frontier, particularly due to the substantial memorization capabilities of DMs that pose significant privacy risks. Differential privacy offers a rigorous framework for safeguarding individual data points during model training, with Differential Privacy Stochastic Gradient Descent (DP-SGD) being a prominent implementation. Diffusion method decomposes image generation into iterative steps, theoretically aligning well with DP's incremental noise addition. Despite the natural fit, the unique architecture of DMs necessitates tailored approaches to effectively balance privacy-utility trade-off. Recent developments in this field have highlighted the potential for generating high-quality synthetic data by pre-training on public data (i.e., ImageNet) and fine-tuning on private data, however, there is a pronounced gap in research on optimizing the trade-offs involved in DP settings, particularly concerning parameter efficiency and model scalability. Our work addresses this by proposing a parameter-efficient fine-tuning strategy optimized for private diffusion models, which minimizes the number of trainable parameters to enhance the privacy-utility trade-off. We empirically demonstrate that our method achieves state-of-the-art performance in DP synthesis, significantly surpassing previous benchmarks on widely studied datasets (e.g., with only 0.47M trainable parameters, achieving a more than 35% improvement over the previous state-of-the-art with a small privacy budget on the CelebA-64 dataset). Anonymous codes available at https://anonymous.4open.science/r/DP-LORA-F02F., Comment: 16 pages, 5 figures, 11 tables

The advent of personalized content generation by LLMs presents a novel challenge: how to efficiently adapt text to meet individual preferences without the unsustainable demand of creating a unique model for each user. This study introduces an innovative online method that employs neural bandit algorithms to dynamically optimize soft instruction embeddings based on user feedback, enhancing the personalization of open-ended text generation by white-box LLMs. Through rigorous experimentation on various tasks, we demonstrate significant performance improvements over baseline strategies. NeuralTS, in particular, leads to substantial enhancements in personalized news headline generation, achieving up to a 62.9% improvement in terms of best ROUGE scores and up to 2.76% increase in LLM-agent evaluation against the baseline., Comment: 7 pages

Machine Learning as a Service (MLaaS) is a popular cloud-based solution for customers who aim to use an ML model but lack training data, computation resources, or expertise in ML. In this case, the training datasets are typically a private possession of the ML or data companies and are inaccessible to the customers, but the customers still need an approach to confirm that the training datasets meet their expectations and fulfil regulatory measures like fairness. However, no existing work addresses the above customers' concerns. This work is the first attempt to solve this problem, taking data origin as an entry point. We first define origin membership measurement and based on this, we then define diversity and fairness metrics to address customers' concerns. We then propose a strategy to estimate the values of these two metrics in the inaccessible training dataset, combining shadow training techniques from membership inference and an efficient featurization scheme in multiple instance learning. The evaluation contains an application of text review polarity classification applications based on the language BERT model. Experimental results show that our solution can achieve up to 0.87 accuracy for membership inspection and up to 99.3% confidence in inspecting diversity and fairness distribution.

LoRaWAN deployments follow an ad-hoc deployment model that has organically led to overlapping communication networks, sharing the wireless spectrum, and completely unaware of each other. LoRaWAN uses ALOHA-style communication where it is almost impossible to schedule transmission between networks belonging to different owners properly. The inability to schedule overlapping networks will cause inter-network interference, which will increase node-to-gateway message losses and gateway-to-node acknowledgement failures. This problem is likely to get worse as the number of LoRaWAN networks increase. In response to this problem, we propose IRONWAN, a wireless overlay network that shares communication resources without modifications to underlying protocols. It utilises the broadcast nature of radio communication and enables gateway-to-gateway communication to facilitate the search for failed messages and transmit failed acknowledgements already received and cached in overlapping network's gateways. IRONWAN uses two novel algorithms, a Real-time Message Inter-arrival Predictor, to highlight when a server has not received an expected uplink message. The Interference Predictor ensures that extra gateway-to-gateway communication does not negatively impact communication bandwidth. We evaluate IRONWAN on a 1000-node simulator with up to ten gateways and a 10-node testbed with 2-gateways. Results show that IRONWAN can achieve up to 12\% higher packet delivery ratio (PDR) and total messages received per node while increasing the minimum PDR by up to 28\%. These improvements save up to 50\% node's energy. Finally, we demonstrate that IRONWAN has comparable performance to an optimal solution (wired, centralised) but with 2-32 times lower communication costs. IRONWAN also has up to 14\% better PDR when compared to FLIP, a wired-distributed gateway-to-gateway protocol in certain scenarios., Comment: Accepted for publication at IEEE Internet of Things Journal

Owing to the complicated characteristics of 5G communication system, designing RF components through mathematical modeling becomes a challenging obstacle. Moreover, such mathematical models need numerous manual adjustments for various specification requirements. In this paper, we present a learning-based framework to model and compensate Power Amplifiers (PAs) in 5G communication. In the proposed framework, Deep Neural Networks (DNNs) are used to learn the characteristics of the PAs, while, correspondent Digital Pre-Distortions (DPDs) are also learned to compensate for the nonlinear and memory effects of PAs. On top of the framework, we further propose two frequency domain losses to guide the learning process to better optimize the target, compared to naive time domain Mean Square Error (MSE). The proposed framework serves as a drop-in replacement for the conventional approach. The proposed approach achieves an average of 56.7% reduction of nonlinear and memory effects, which converts to an average of 16.3% improvement over a carefully-designed mathematical model, and even reaches 34% enhancement in severe distortion scenarios., Comment: IEEE International Conference on Communications (ICC) 2021

LoRaWAN is one of the leading Low Power Wide Area Network (LPWAN) architectures. It was originally designed for systems consisting of static sensor or Internet of Things (IoT) devices and static gateways. It was recently updated to introduce new features such as nano-second timestamps which open up applications to enable LoRaWAN to be adopted for mobile device tracking and localisation. In such mobile scenarios, devices could temporarily lose communication with the gateways because of interference from obstacles or deep fading, causing throughput reduction and delays in data transmission. To overcome this problem, we propose a new data forwarding scheme. Instead of holding the data until the next contact with gateways, devices can forward their data to nearby devices that have a higher probability of being in contact with gateways. We propose a new network metric called Real-Time Contact-Aware Expected Transmission Count (RCA-ETX) to model this contact probability in real-time. Without making any assumption on mobility models, this metric exploits data transmission delays to model complex device mobility. We also extend RCA-ETX with a throughput-optimal stochastic backpressure routing scheme and propose Real-Time Opportunistic Backpressure Collection (ROBC), a protocol to counter the stochastic behaviours resulting from the dynamics associated with mobility. To apply our approaches seamlessly to LoRaWAN-enabled devices, we further propose two new LaRaWAN classes, namely Modified Class-C and Queue-based Class-A. Both of them are compatible with LoRaWAN Class-A devices. Our data-driven experiments, based on the London bus network, show that our approaches can reduce data transmission delays up to $25\%$ and provide a $53\%$ throughput improvement in data transfer performance., Comment: Accepted for publication at ICDCS 2020

Wireless Sensor Network (WSN) applications range from domestic Internet of Things systems like temperature monitoring of homes to the monitoring and control of large-scale critical infrastructures. The greatest risk with the use of WSNs in critical infrastructure is their vulnerability to malicious network level attacks. Their radio communication network can be disrupted, causing them to lose or delay data which will compromise system functionality. This paper presents Antilizer, a lightweight, fully-distributed solution to enable WSNs to detect and recover from common network level attack scenarios. In Antilizer each sensor node builds a self-referenced trust model of its neighbourhood using network overhearing. The node uses the trust model to autonomously adapt its communication decisions. In the case of a network attack, a node can make neighbour collaboration routing decisions to avoid affected regions of the network. Mobile agents further bound the damage caused by attacks. These agents enable a simple notification scheme which propagates collaborative decisions from the nodes to the base station. A filtering mechanism at the base station further validates the authenticity of the information shared by mobile agents. We evaluate Antilizer in simulation against several routing attacks. Our results show that Antilizer reduces data loss down to 1% (4% on average), with operational overheads of less than 1% and provides fast network-wide convergence.