Investigation of the degree of personalization in federated learning algorithms has shown that only maximizing the performance of the global model will confine the capacity of the local models to personalize. In this paper, we advocate an adaptive personalized federated learning (APFL) algorithm, where each client will train their local models while contributing to the global model. Theoretically, we show that the mixture of local and global models can reduce the generalization error, using the multi-domain learning theory. We also propose a communication-reduced bilevel optimization method, which reduces the communication rounds to $O(\sqrt{T})$ and show that under strong convexity and smoothness assumptions, the proposed algorithm can achieve a convergence rate of $O(1/T)$ with some residual error. The residual error is related to the gradient diversity among local models, and the gap between optimal local and global models.

Reverse Vending Machines (RVMs) are a proven instrument for facilitating closed-loop plastic packaging recycling. A good customer experience at the RVM is crucial for a further proliferation of this technology. Bin full events are the major reason for Reverse Vending Machine (RVM) downtime at the world leader in the RVM market. The paper at hand develops and evaluates an approach based on machine learning and statistical approximation to foresee bin full events and, thus increase uptime of RVMs. Our approach relies on forecasting the hourly time series of returned beverage containers at a given RVM. We contribute by developing and evaluating an approach for hourly forecasts in a retail setting - this combination of application domain and forecast granularity is novel. A trace-driven simulation confirms that the forecasting-based approach leads to less downtime and costs than naive emptying strategies.

Distance Metric Learning (DML) has drawn much attention over the last two decades. A number of previous works have shown that it performs well in measuring the similarities of individuals given a set of correctly labeled pairwise data by domain experts. These important and precisely-labeled pairwise data are often highly sensitive in real world (e.g., patients similarity). This paper studies, for the first time, how pairwise information can be leaked to attackers during distance metric learning, and develops differential pairwise privacy (DPP), generalizing the definition of standard differential privacy, for secure metric learning. Unlike traditional differential privacy which only applies to independent samples, thus cannot be used for pairwise data, DPP successfully deals with this problem by reformulating the worst case. Specifically, given the pairwise data, we reveal all the involved correlations among pairs in the constructed undirected graph. DPP is then formalized that defines what kind of DML algorithm is private to preserve pairwise data. After that, a case study employing the contrastive loss is exhibited to clarify the details of implementing a DPP-DML algorithm. Particularly, the sensitivity reduction technique is proposed to enhance the utility of the output distance metric. Experiments both on a toy dataset and benchmarks demonstrate that the proposed scheme achieves pairwise data privacy without compromising the output performance much (Accuracy declines less than 0.01 throughout all benchmark datasets when the privacy budget is set at 4).

Parameter space exploration methods with black-box optimization have recently been shown to outperform state-of-the-art approaches in continuous control reinforcement learning domains. In this paper, we examine reasons why these methods work better and the situations in which they are worse than traditional action space exploration methods. Through a simple theoretical analysis, we show that when the parametric complexity required to solve the reinforcement learning problem is greater than the product of action space dimensionality and horizon length, exploration in action space is preferred. This is also shown empirically by comparing simple exploration methods on several toy problems.

Graph representation learning (GRL) is a powerful technique for learning low-dimensional vector representation of high-dimensional and often sparse graphs. Most studies explore the structure and metadata associated with the graph using random walks and employ an unsupervised or semi-supervised learning schemes. Learning in these methods is context-free, resulting in only a single representation per node. Recently studies have argued on the adequacy of a single representation and proposed context-sensitive approaches, which are capable of extracting multiple node representations for different contexts. This proved to be highly effective in applications such as link prediction and ranking. However, most of these methods rely on additional textual features that require complex and expensive RNNs or CNNs to capture high-level features or rely on a community detection algorithm to identify multiple contexts of a node. In this study we show that in-order to extract high-quality context-sensitive node representations it is not needed to rely on supplementary node features, nor to employ computationally heavy and complex models. We propose GOAT, a context-sensitive algorithm inspired by gossip communication and a mutual attention mechanism simply over the structure of the graph. We show the efficacy of GOAT using 6 real-world datasets on link prediction and node clustering tasks and compare it against 12 popular and state-of-the-art (SOTA) baselines. GOAT consistently outperforms them and achieves up to 12% and 19% gain over the best performing methods on link prediction and clustering tasks, respectively.

Topic Detection and Tracking (TDT) is a very active research question within the area of text mining, generally applied to news feeds and Twitter datasets, where topics and events are detected. The notion of "event" is broad, but typically it applies to occurrences that can be detected from a single post or a message. Little attention has been drawn to what we call "micro-events", which, due to their nature, cannot be detected from a single piece of textual information. The study investigates micro-event detection on textual data using a sample of messages from the Stack Overflow Q&A platform in order to detect Free/Libre Open Source Software (FLOSS) version releases. Micro-events are detected using logistic regression models with step-wise forward regression feature selection from a set of LDA topics and sentiment analysis features. We perform a detailed statistical analysis of the models, including influential cases, variance inflation factors, validation of the linearity assumption, pseudo R squared measures and no-information rate. Finally, in order to understand the detection limits and improve the performance of the estimators, we suggest a method for generating micro-event synthetic datasets and use them identify the micro-event detectability thresholds.

In this paper we analyze the necessary number of samples to estimate the gradient of any multidimensional smooth (possibly non-convex) function in a zero-order stochastic oracle model. In this model, an estimator has access to noisy values of the function, in order to produce the estimate of the gradient. We also provide an analysis on the sufficient number of samples for the finite difference method, a classical technique in numerical linear algebra. For $T$ samples and $d$ dimensions, our information-theoretic lower bound is $\Omega(\sqrt{d/T})$. We show that the finite difference method has rate $O(d^{4/3}/\sqrt{T})$ for functions with zero third and higher order derivatives. Thus, the finite difference method is not minimax optimal, and therefore there is space for the development of better gradient estimation methods.

Choosing a deep neural network architecture is a fundamental problem in applications that require balancing performance and parameter efficiency. Standard approaches rely on ad-hoc engineering or computationally expensive validation on a specific dataset. We instead attempt to quantify networks by their intrinsic capacity for unique and robust representations, enabling efficient architecture comparisons without requiring any data. Building upon theoretical connections between deep learning and sparse approximation, we propose the deep frame potential: a measure of coherence that is approximately related to representation stability but has minimizers that depend only on network structure. This provides a framework for jointly quantifying the contributions of architectural hyper-parameters such as depth, width, and skip connections. We validate its use as a criterion for model selection and demonstrate correlation with generalization error on a variety of common residual and densely connected network architectures.

Federated learning is a machine learning setting where a set of edge devices iteratively train a model under the orchestration of a central server, while keeping all data locally on edge devices. In each iteration of federated learning, edge devices perform computation with their local data, and the local computation results are then uploaded to the server for model update. During this process, the challenges of privacy leakage and communication overhead arise due to the extensive information exchange between edge devices and the server. In this paper, we develop CPFed, a communication-efficient and privacy-preserving federated learning method, to solve the above challenges. CPFed integrates three key components: (1) periodic averaging where local computation results at edge devices are only periodically averaged at the server; (2) Gaussian mechanism where edge devices randomly perturb their local computation results before sending the results to the server; and (3) secure aggregation where the perturbed local computation results are homomorphically encrypted before being sent to the server. CPFed can address both the communication efficiency and privacy leakage challenges in federated learning while achieving high model accuracy. We provide an end-to-end privacy guarantee of CPFed and analyze its theoretical convergence rates for both convex and non-convex models. Through extensive numerical experiments on real-world datasets, we demonstrate the effectiveness and efficiency of our proposed method.

We propose a novel regularization-based continual learning method, dubbed as Adaptive Group Sparsity based Continual Learning (AGS-CL), using two group sparsity-based penalties. Our method selectively employs the two penalties when learning each node based its the importance, which is adaptively updated after learning each new task. By utilizing the proximal gradient descent method for learning, the exact sparsity and freezing of the model is guaranteed, and thus, the learner can explicitly control the model capacity as the learning continues. Furthermore, as a critical detail, we re-initialize the weights associated with unimportant nodes after learning each task in order to prevent the negative transfer that causes the catastrophic forgetting and facilitate efficient learning of new tasks. Throughout the extensive experimental results, we show that our AGS-CL uses much less additional memory space for storing the regularization parameters, and it significantly outperforms several state-of-the-art baselines on representative continual learning benchmarks for both supervised and reinforcement learning tasks.