The rise of generative APIs has fueled interest in privacy-preserving synthetic data generation. While the Private Evolution (PE) algorithm generates Differential Privacy (DP) synthetic images using diffusion model APIs, it struggles with few-shot private data due to the limitations of its DP-protected similarity voting approach. In practice, the few-shot private data challenge is particularly prevalent in specialized domains like healthcare and industry. To address this challenge, we propose a novel API-assisted algorithm, Private Contrastive Evolution (PCEvolve), which iteratively mines inherent inter-class contrastive relationships in few-shot private data beyond individual data points and seamlessly integrates them into an adapted Exponential Mechanism (EM) to optimize DP’s utility in an evolution loop. We conduct extensive experiments on four specialized datasets, demonstrating that PCEvolve outperforms PE and other API-assisted baselines. These results highlight the potential of leveraging API access with private data for quality evaluation, enabling the generation of high-quality DP synthetic images and paving the way for more accessible and effective privacy-preserving generative API applications. Our code is available at https://github.com/TsingZ0/PCEvolve.

Substantial quantity and high quality are the golden rules of making a good training dataset with sample privacy protection equally important. Generating synthetic samples that resemble high-quality private data while ensuring Differential Privacy (DP), a formal privacy guarantee, promises scalability and practicality. However, existing methods relying on pre-trained models for data synthesis often struggle in data-deficient scenarios, suffering from limited sample size, inevitable generation noise and existing pre-trained model bias. To address these challenges, we propose a novel contr**A**stive private data **S**ynthesis via **W**eighted multiple **P**re-trained generative models framework, named as **WASP**. WASP utilizes limited private samples for more accurate private data distribution estimation via a Top-*Q* voting mechanism, and leverages low-quality synthetic samples for contrastive generation via collaboration among dynamically weighted multiple pre-trained models. Extensive experiments on 6 well-developed datasets with 6 open-source and 3 closed-source PLMs demonstrate the superiority of WASP in improving model performance over diverse downstream tasks. Code is available at https://github.com/LindaLydia/WASP.

Vertical Federated Learning (VFL) is widely utilized in real-world applications to enable collaborative learning while protecting local data and models. However, previous works show that parties without labels (passive parties) in VFL can infer the sensitive label or feature information owned by the party with labels (active party), or execute backdoor attacks. Meanwhile, active party can also infer sensitive feature or attribute information from passive party. All these pose great challenges to VFL systems. Former defense methods tend to experience either a loss in overall effectiveness or are too specialized for specific tasks. In this work, we propose a novel method and a unified framework for defending various attacks in VFL altogether, namely Mutual Information Regularization Defense (MID), which limits the mutual information between private raw data and intermediate outputs to achieve a consistently better trade-off between model utility and privacy. We provide both theoretical and experimental evidence to confirm the effectiveness of our MID framework in defending against a wide range of label and feature inference attacks, along with backdoor attacks in VFL. These showcase its promising potential as a versatile and effective defense mechanism, not tied to any specific task.

Data-generation based zero-shot learning, although effective in training Small Task-specific Models (STMs) via synthetic datasets generated by Pre-trained Language Models (PLMs), is often limited by the low quality of such synthetic datasets.Previous solutions have primarily focused on single PLM settings, where synthetic datasets are typically restricted to specific sub-spaces and often deviate from real-world distributions, leading to severe distribution bias.To mitigate such bias, we propose FuseGen, a novel data-generation based zero-shot learning framework that introduces a new criteria for subset selection from synthetic datasets via utilizing multiple PLMs and trained STMs.The chosen subset provides in-context feedback to each PLM, enhancing dataset quality through iterative data generation.Trained STMs are then used for sample re-weighting as well, further improving data quality.Extensive experiments across diverse tasks demonstrate that FuseGen substantially outperforms existing methods, highly effective in boosting STM performance in a PLM-agnostic way. 1

Vertical Federated Learning (VFL) has emerged as a collaborative training paradigm that allows participants with different features of the same group of users to accomplish cooperative training without exposing their raw data or model parameters. VFL has gained significant attention for its research potential and real-world applications in recent years, but still faces substantial challenges, such as in defending various kinds of data inference and backdoor attacks. Moreover, most of existing VFL projects are industry-facing and not easily used for keeping track of the current research progress. To address this need, we present an extensible and lightweight VFL framework VFLAIR (available at https://github.com/FLAIR-THU/VFLAIR), which supports VFL training with a variety of models, datasets and protocols, along with standardized modules for comprehensive evaluations of attacks and defense strategies. We also benchmark $11$ attacks and $8$ defenses performance under different communication and model partition settings and draw concrete insights and recommendations on the choice of defense strategies for different practical VFL deployment scenarios.

In a vertical federated learning (VFL) scenario where features and models are split into different parties, it has been shown that sample-level gradient information can be exploited to deduce crucial label information that should be kept secret. An immediate defense strategy is to protect sample-level messages communicated with Homomorphic Encryption (HE), exposing only batch-averaged local gradients to each party. In this paper, we show that even with HE-protected communication, private labels can still be reconstructed with high accuracy by gradient inversion attack, contrary to the common belief that batch-averaged information is safe to share under encryption. We then show that backdoor attack can also be conducted by directly replacing encrypted communicated messages without decryption. To tackle these attacks, we propose a novel defense method, Confusional AutoEncoder (termedCAE), which is based on autoencoder and entropy regularization to disguise true labels. To further defend attackers with sufficient prior label knowledge, we introduce DiscreteSGD-enhanced CAE (termedDCAE), and show that DCAE significantly boosts the main task accuracy than other known methods when defending various label inference attacks.

Vertical Federated Learning (VFL) is a federated learning setting where multiple parties with different features about the same set of users jointly train machine learning models without exposing their raw data or model parameters. Motivated by the rapid growth in VFL research and real-world applications, we provide a comprehensive review of the concept and algorithms of VFL, as well as current advances and challenges in various aspects, including effectiveness, efficiency, and privacy. We provide an exhaustive categorization for VFL settings and privacy-preserving protocols and comprehensively analyze the privacy attacks and defense strategies for each protocol. In the end, we propose a unified framework, termed VFLow, which considers the VFL problem under communication, computation, privacy, as well as effectiveness and fairness constraints. Finally, we review the most recent advances in industrial applications, highlighting open challenges and future directions for VFL.

Smartphone overuse is related to a variety of issues such as lack of sleep and anxiety. We explore the application of Self-Affirmation Theory on smartphone overuse intervention in a just-in-time manner. We present TypeOut, a just-in-time intervention technique that integrates two components: an in-situ typing-based unlock process to improve user engagement, and self-affirmation-based typing content to enhance effectiveness. We hypothesize that the integration of typing and self-affirmation content can better reduce smartphone overuse. We conducted a 10-week within-subject field experiment (N=54) and compared TypeOut against two baselines: one only showing the self-affirmation content (a common notification-based intervention), and one only requiring typing non-semantic content (a state-of-the-art method). TypeOut reduces app usage by over 50%, and both app opening frequency and usage duration by over 25%, all significantly outperforming baselines. TypeOut can potentially be used in other domains where an intervention may benefit from integrating self-affirmation exercises with an engaging just-in-time mechanism.

Learning to write is challenging for blind and low vision (BLV) people because of the lack of visual feedback. Regardless of the drastic advancement of digital technology, handwriting is still an essential part of daily life. Although tools designed for teaching BLV to write exist, many are expensive and require the help of sighted teachers. We propose LightWrite, a low-cost, easy-to-access smartphone application that uses voice-based descriptive instruction and feedback to teach BLV users to write English lowercase letters and Arabian digits in a specifically designed font. A two-stage study with 15 BLV users with little prior writing knowledge shows that LightWrite can successfully teach users to learn handwriting characters in an average of 1.09 minutes for each letter. After initial training and 20-minute daily practice for 5 days, participants were able to write an average of 19.9 out of 26 letters that are recognizable by sighted raters.