How can we find a subset of training samples that are most responsible for a complicated black-box machine learning model prediction? More generally, how can we explain the model decision to end-users in a transparent way? We propose a new model-agnostic algorithm to identify a minimum number of training samples that are indispensable for a given model decision at a particular test point, as the model decision would otherwise change upon the removal of these training samples. In line with the counterfactual explanation, our algorithm identifies such a set of indispensable samples iteratively by solving a constrained optimization problem. Further, we efficiently speed up the algorithm through approximation. To demonstrate the effectiveness of the algorithm, we apply it to a variety of tasks including data poisoning detection, training set debugging, and understanding loan decisions. Results show that our algorithm is an effective and easy to comprehend tool to help better understand local model behaviors and therefore facilitate the application of machine learning in domains where such understanding is a requisite and where end-users do not have a machine learning background.
Graph neural networks (GNNs) aim to learn graph representations that preserve both attributive and structural information. In this paper, we study the problem of how to select high-quality nodes for training GNNs, considering GNNs are sensitive to different training datasets. Active learning (AL), whose purpose is to find the most informative instances to maximize the performance of the model, is a promising approach to solve this problem. Previous attempts have combined AL with graph representation learning by designing several selection criteria to measure how informative a node is. However, these methods do not directly utilize both the rich semantic and structural information and are prone to select sparsely-connected nodes (i.e. nodes having few neighbors) and low-purity nodes (i.e. nodes having noisy inter-class edges), which are less effective for training GNN models. To address these problems, we present a Deep Active Graph Representation Learning framework (DAGRL), in which three novel selection criteria are proposed. Specifically, we propose to measure the uncertainty of nodes via random topological perturbation. Besides, we propose two novel representativeness sampling criteria, which utilize both the structural and label information to find densely-connected nodes with many intra-class edges, hence enhance the performance of GNN models significantly. Then, we combine these three criteria with time-sensitive scheduling in accordance to the training progress of GNNs. Furthermore, considering the different size of classes, we employ a novel cluster-aware node selection policy, which ensures the number of selected nodes in each class is proportional to the size of the class. Comprehensive experiments on three public datasets show that our method outperforms previous baselines by a significant margin, which demonstrates its effectiveness.
Despite the great success of deep learning, recent works show that large deep neural networks are often highly redundant and can be significantly reduced in size. However, the theoretical question of how much we can prune a neural network given a specified tolerance of accuracy drop is still open. This paper provides one answer to this question by proposing a greedy optimization based pruning method. The proposed method has the guarantee that the discrepancy between the pruned network and the original network decays with exponentially fast rate w.r.t. the size of the pruned network, under weak assumptions that apply for most practical settings. Empirically, our method improves prior arts on pruning various network architectures including ResNet, MobilenetV2/V3 on ImageNet.
Off-policy evaluation provides an essential tool for evaluating the effects of different policies or treatments using only observed data. When applied to high-stakes scenarios such as medical diagnosis or financial decision-making, it is crucial to provide provably correct upper and lower bounds of the expected reward, not just a classical single point estimate, to the end-users, as executing a poor policy can be very costly. In this work, we propose a provably correct method for obtaining interval bounds for off-policy evaluation in a general continuous setting. The idea is to search for the maximum and minimum values of the expected reward among all the Lipschitz Q-functions that are consistent with the observations, which amounts to solving a constrained optimization problem on a Lipschitz function space. We go on to introduce a Lipschitz value iteration method to monotonically tighten the interval, which is simple yet efficient and provably convergent. We demonstrate the practical efficiency of our method on a range of benchmarks.
Recently, contrastive learning (CL) has emerged as a successful method for unsupervised graph representation learning. Most graph CL methods first perform stochastic augmentation on the input graph to obtain two graph views and maximize the agreement of representations in the two views. Despite the prosperous development of graph CL methods, the design of graph augmentation schemes---a crucial component in CL---remains rarely explored. We argue that the data augmentation schemes should preserve intrinsic structural and attribute information of graphs, which will force the model to learn representations that are insensitive to perturbation on unimportant nodes and edges. However, most existing methods adopt uniform data augmentation schemes, like uniformly dropping edges and uniformly shuffling features, leading to suboptimal performance. In this paper, we propose a novel graph contrastive representation learning method with adaptive augmentation that incorporates various priors for topological and semantic aspects of the graph. Specifically, on the topology level, we design augmentation schemes based on node centrality measures to highlight important connective structures. On the node attribute level, we corrupt node features by adding more noise to unimportant node features, to enforce the model to recognize underlying semantic information. We perform extensive experiments of node classification on a variety of real-world datasets. Experimental results demonstrate that our proposed method consistently outperforms existing state-of-the-art methods and even surpasses some supervised counterparts, which validates the effectiveness of the proposed contrastive framework with adaptive augmentation.
In this work we addressed the problem of capturing sequential information contained in longitudinal electronic health records (EHRs). Clinical notes, which is a particular type of EHR data, are a rich source of information and practitioners often develop clever solutions how to maximise the sequential information contained in free-texts. We proposed a systematic methodology for learning from chronological events available in clinical notes. The proposed methodological {\it path signature} framework creates a non-parametric hierarchical representation of sequential events of any type and can be used as features for downstream statistical learning tasks. The methodology was developed and externally validated using the largest in the UK secondary care mental health EHR data on a specific task of predicting survival risk of patients diagnosed with Alzheimer's disease. The signature-based model was compared to a common survival random forest model. Our results showed a 15.4$\%$ increase of risk prediction AUC at the time point of 20 months after the first admission to a specialist memory clinic and the signature method outperformed the baseline mixed-effects model by 13.2 $\%$.
The most successful methods such as ReLU transfer functions, batch normalization, Xavier initialization, dropout, learning rate decay, or dynamic optimizers, have become standards in the field due, particularly, to their ability to increase the performance of Neural Networks (NNs) significantly and in almost all situations. Here we present a new method to calculate the gradients while training NNs, and show that it significantly improves final performance across architectures, data-sets, hyper-parameter values, training length, and model sizes, including when it is being combined with other common performance-improving methods (such as the ones mentioned above). Besides being effective in the wide array situations that we have tested, the increase in performance (e.g. F1) it provides is as high or higher than this one of all the other widespread performance-improving methods that we have compared against. We call our method Population Gradients (PG), and it consists on using a population of NNs to calculate a non-local estimation of the gradient, which is closer to the theoretical exact gradient (i.e. this one obtainable only with an infinitely big data-set) of the error function than the empirical gradient (i.e. this one obtained with the real finite data-set).
Large scale deep learning provides a tremendous opportunity to improve the quality of content recommendation systems by employing both wider and deeper models, but this comes at great infrastructural cost and carbon footprint in modern data centers. Pruning is an effective technique that reduces both memory and compute demand for model inference. However, pruning for online recommendation systems is challenging due to the continuous data distribution shift (a.k.a non-stationary data). Although incremental training on the full model is able to adapt to the non-stationary data, directly applying it on the pruned model leads to accuracy loss. This is because the sparsity pattern after pruning requires adjustment to learn new patterns. To the best of our knowledge, this is the first work to provide in-depth analysis and discussion of applying pruning to online recommendation systems with non-stationary data distribution. Overall, this work makes the following contributions: 1) We present an adaptive dense to sparse paradigm equipped with a novel pruning algorithm for pruning a large scale recommendation system with non-stationary data distribution; 2) We design the pruning algorithm to automatically learn the sparsity across layers to avoid repeating hand-tuning, which is critical for pruning the heterogeneous architectures of recommendation systems trained with non-stationary data.