Federated learning (FL) has gained significant attention recently as a privacy-enhancing tool to jointly train a machine learning model by multiple participants. The prior work on FL has mostly studied how to protect label privacy during model training. However, model evaluation in FL might also lead to potential leakage of private label information. In this work, we propose an evaluation algorithm that can accurately compute the widely used AUC (area under the curve) metric when using the label differential privacy (DP) in FL. Through extensive experiments, we show our algorithms can compute accurate AUCs compared to the ground truth. The code is available at {\url{https://github.com/bytedance/fedlearner/tree/master/example/privacy/DPAUC}}.

Online streaming feature selection (OSFS), which conducts feature selection in an online manner, plays an important role in dealing with high-dimensional data. In many real applications such as intelligent healthcare platform, streaming feature always has some missing data, which raises a crucial challenge in conducting OSFS, i.e., how to establish the uncertain relationship between sparse streaming features and labels. Unfortunately, existing OSFS algorithms never consider such uncertain relationship. To fill this gap, we in this paper propose an online sparse streaming feature selection with uncertainty (OS2FSU) algorithm. OS2FSU consists of two main parts: 1) latent factor analysis is utilized to pre-estimate the missing data in sparse streaming features before con-ducting feature selection, and 2) fuzzy logic and neighborhood rough set are employed to alleviate the uncertainty between estimated streaming features and labels during conducting feature selection. In the experiments, OS2FSU is compared with five state-of-the-art OSFS algorithms on six real datasets. The results demonstrate that OS2FSU outperforms its competitors when missing data are encountered in OSFS.

Federated learning (FL) is a privacy-preserving distributed machine learning technique that trains models while keeping all the original data generated on devices locally. Since devices may be resource constrained, offloading can be used to improve FL performance by transferring computational workload from devices to edge servers. However, due to mobility, devices participating in FL may leave the network during training and need to connect to a different edge server. This is challenging because the offloaded computations from edge server need to be migrated. In line with this assertion, we present FedFly, which is, to the best of our knowledge, the first work to migrate a deep neural network (DNN) when devices move between edge servers during FL training. Our empirical results on the CIFAR10 dataset, with both balanced and imbalanced data distribution, support our claims that FedFly can reduce training time by up to 33% when a device moves after 50% of the training is completed, and by up to 45% when 90% of the training is completed when compared to state-of-the-art offloading approach in FL. FedFly has negligible overhead of up to two seconds and does not compromise accuracy. Finally, we highlight a number of open research issues for further investigation. FedFly can be downloaded from https://github.com/qub-blesson/FedFly.

Learning from multiple related tasks by knowledge sharing and transfer has become increasingly relevant over the last two decades. In order to successfully transfer information from one task to another, it is critical to understand the similarities and differences between the domains. In this paper, we introduce the notion of \emph{performance gap}, an intuitive and novel measure of the distance between learning tasks. Unlike existing measures which are used as tools to bound the difference of expected risks between tasks (e.g., $\mathcal{H}$-divergence or discrepancy distance), we theoretically show that the performance gap can be viewed as a data- and algorithm-dependent regularizer, which controls the model complexity and leads to finer guarantees. More importantly, it also provides new insights and motivates a novel principle for designing strategies for knowledge sharing and transfer: gap minimization. We instantiate this principle with two algorithms: 1. {gapBoost}, a novel and principled boosting algorithm that explicitly minimizes the performance gap between source and target domains for transfer learning; and 2. {gapMTNN}, a representation learning algorithm that reformulates gap minimization as semantic conditional matching for multitask learning. Our extensive evaluation on both transfer learning and multitask learning benchmark data sets shows that our methods outperform existing baselines.

Recently unsupervised representation learning (URL) has achieved remarkable progress in various scenarios. However, most methods are specifically designed based on specific data characters or task assumptions. Based on the manifold assumption, we regard most URL problems as an embedding problem that seeks an optimal low-dimensional representation of the given high-dimensional data. We split the embedding process into two steps, data structural modeling and low-dimensional embedding, and propose a general similarity-based framework called GenURL. Specifically, we provide a general method to model data structures by adaptively combining graph distances on the feature space and predefined graphs, then propose robust loss functions to learn the low-dimensional embedding. Combining with a specific pretext task, we can adapt GenURL to various URL tasks in a unified manner and achieve state-of-the-art performance, including self-supervised visual representation learning, unsupervised knowledge distillation, graph embeddings, and dimension reduction. Moreover, ablation studies of loss functions and basic hyper-parameter settings in GenURL illustrate the data characters of various tasks.

Recent development in brain-machine interface technology has made seizure prediction possible. However, the communication of large volume of electrophysiological signals between sensors and processing apparatus and related computation become two major bottlenecks for seizure prediction systems due to the constrained bandwidth and limited computation resource, especially for wearable and implantable medical devices. Although compressive sensing (CS) can be adopted to compress the signals to reduce communication bandwidth requirement, it needs a complex reconstruction procedure before the signal can be used for seizure prediction. In this paper, we propose C$^2$SP-Net, to jointly solve compression, prediction, and reconstruction with a single neural network. A plug-and-play in-sensor compression matrix is constructed to reduce transmission bandwidth requirement. The compressed signal can be used for seizure prediction without additional reconstruction steps. Reconstruction of the original signal can also be carried out in high fidelity. Prediction accuracy, sensitivity, false prediction rate, and reconstruction quality of the proposed framework are evaluated under various compression ratios. The experimental results illustrate that our model outperforms the competitive state-of-the-art baselines by a large margin in prediction accuracy. In particular, our proposed method produces an average loss of 0.35 % in prediction accuracy with a compression ratio ranging from 1/2 to 1/16.

This paper describes the University of Sydney& JD's joint submission of the IWSLT 2021 low resource speech translation task. We participated in the Swahili-English direction and got the best scareBLEU (25.3) score among all the participants. Our constrained system is based on a pipeline framework, i.e. ASR and NMT. We trained our models with the officially provided ASR and MT datasets. The ASR system is based on the open-sourced tool Kaldi and this work mainly explores how to make the most of the NMT models. To reduce the punctuation errors generated by the ASR model, we employ our previous work SlotRefine to train a punctuation correction model. To achieve better translation performance, we explored the most recent effective strategies, including back translation, knowledge distillation, multi-feature reranking and transductive finetuning. For model structure, we tried auto-regressive and non-autoregressive models, respectively. In addition, we proposed two novel pre-train approaches, i.e. \textit{de-noising training} and \textit{bidirectional training} to fully exploit the data. Extensive experiments show that adding the above techniques consistently improves the BLEU scores, and the final submission system outperforms the baseline (Transformer ensemble model trained with the original parallel data) by approximately 10.8 BLEU score, achieving the SOTA performance.

Unsupervised attributed graph representation learning is challenging since both structural and feature information are required to be represented in the latent space. Existing methods concentrate on learning latent representation via reconstruction tasks, but cannot directly optimize representation and are prone to oversmoothing, thus limiting the applications on downstream tasks. To alleviate these issues, we propose a novel graph embedding framework named Deep Manifold Attributed Graph Embedding (DMAGE). A node-to-node geodesic similarity is proposed to compute the inter-node similarity between the data space and the latent space and then use Bergman divergence as loss function to minimize the difference between them. We then design a new network structure with fewer aggregation to alleviate the oversmoothing problem and incorporate graph structure augmentation to improve the representation's stability. Our proposed DMAGE surpasses state-of-the-art methods by a significant margin on three downstream tasks: unsupervised visualization, node clustering, and link prediction across four popular datasets.

Spoken language understanding (SLU) system usually consists of various pipeline components, where each component heavily relies on the results of its upstream ones. For example, Intent detection (ID), and slot filling (SF) require its upstream automatic speech recognition (ASR) to transform the voice into text. In this case, the upstream perturbations, e.g. ASR errors, environmental noise and careless user speaking, will propagate to the ID and SF models, thus deteriorating the system performance. Therefore, the well-performing SF and ID models are expected to be noise resistant to some extent. However, existing models are trained on clean data, which causes a \textit{gap between clean data training and real-world inference.} To bridge the gap, we propose a method from the perspective of domain adaptation, by which both high- and low-quality samples are embedding into similar vector space. Meanwhile, we design a denoising generation model to reduce the impact of the low-quality samples. Experiments on the widely-used dataset, i.e. Snips, and large scale in-house dataset (10 million training examples) demonstrate that this method not only outperforms the baseline models on real-world (noisy) corpus but also enhances the robustness, that is, it produces high-quality results under a noisy environment. The source code will be released.

Deep learning has grown rapidly thanks to its state-of-the-art performance across a wide range of real-world applications. While neural networks have been trained using IEEE-754 binary32 arithmetic, the rapid growth of computational demands in deep learning has boosted interest in faster, low precision training. Mixed-precision training that combines IEEE-754 binary16 with IEEE-754 binary32 has been tried, and other $16$-bit formats, for example Google's bfloat16, have become popular. In floating-point arithmetic there is a tradeoff between precision and representation range as the number of exponent bits changes; denormal numbers extend the representation range. This raises questions of how much exponent range is needed, of whether there is a format between binary16 (5 exponent bits) and bfloat16 (8 exponent bits) that works better than either of them, and whether or not denormals are necessary. In the current paper we study the need for denormal numbers for mixed-precision training, and we propose a 1/6/9 format, i.e., 6-bit exponent and 9-bit explicit mantissa, that offers a better range-precision tradeoff. We show that 1/6/9 mixed-precision training is able to speed up training on hardware that incurs a performance slowdown on denormal operations or eliminates the need for denormal numbers altogether. And, for a number of fully connected and convolutional neural networks in computer vision and natural language processing, 1/6/9 achieves numerical parity to standard mixed-precision.