It has long been recognized that academic success is a result of both cognitive and non-cognitive dimensions acting together. Consequently, any intelligent learning platform designed to improve learning outcomes (LOs) must provide actionable inputs to the learner in these dimensions. However, operationalizing such inputs in a production setting that is scalable is not trivial. We develop an Embibe Score Quotient model (ESQ) to predict test scores based on observed academic, behavioral and test-taking features of a student. ESQ can be used to predict the future scoring potential of a student as well as offer personalized learning nudges, both critical to improving LOs. Multiple machine learning models are evaluated for the prediction task. In order to provide meaningful feedback to the learner, individualized Shapley feature attributions for each feature are computed. Prediction intervals are obtained by applying non-parametric quantile regression, in an attempt to quantify the uncertainty in the predictions. We apply the above modelling strategy on a dataset consisting of more than a hundred million learner interactions on the Embibe learning platform. We observe that the Median Absolute Error between the observed and predicted scores is 4.58% across several user segments, and the correlation between predicted and observed responses is 0.93. Game-like what-if scenarios are played out to see the changes in LOs, on counterfactual examples. We briefly discuss how a rational agent can then apply an optimal policy to affect the learning outcomes by treating the above model like an Oracle.

We propose and analyze algorithms for distributionally robust optimization of convex losses with conditional value at risk (CVaR) and $\chi^2$ divergence uncertainty sets. We prove that our algorithms require a number of gradient evaluations independent of training set size and number of parameters, making them suitable for large-scale applications. For $\chi^2$ uncertainty sets these are the first such guarantees in the literature, and for CVaR our guarantees scale linearly in the uncertainty level rather than quadratically as in previous work. We also provide lower bounds proving the worst-case optimality of our algorithms for CVaR and a penalized version of the $\chi^2$ problem. Our primary technical contributions are novel bounds on the bias of batch robust risk estimation and the variance of a multilevel Monte Carlo gradient estimator due to [Blanchet & Glynn, 2015]. Experiments on MNIST and ImageNet confirm the theoretical scaling of our algorithms, which are 9--36 times more efficient than full-batch methods.

Meta-learning aims to perform fast adaptation on a new task through learning a "prior" from multiple existing tasks. A common practice in meta-learning is to perform a train-validation split where the prior adapts to the task on one split of the data, and the resulting predictor is evaluated on another split. Despite its prevalence, the importance of the train-validation split is not well understood either in theory or in practice, particularly in comparison to the more direct non-splitting method, which uses all the per-task data for both training and evaluation. We provide a detailed theoretical study on whether and when the train-validation split is helpful on the linear centroid meta-learning problem, in the asymptotic setting where the number of tasks goes to infinity. We show that the splitting method converges to the optimal prior as expected, whereas the non-splitting method does not in general without structural assumptions on the data. In contrast, if the data are generated from linear models (the realizable regime), we show that both the splitting and non-splitting methods converge to the optimal prior. Further, perhaps surprisingly, our main result shows that the non-splitting method achieves a strictly better asymptotic excess risk under this data distribution, even when the regularization parameter and split ratio are optimally tuned for both methods. Our results highlight that data splitting may not always be preferable, especially when the data is realizable by the model. We validate our theories by experimentally showing that the non-splitting method can indeed outperform the splitting method, on both simulations and real meta-learning tasks.

In Continual Learning, a Neural Network is trained on a stream of data whose distribution shifts over time. Under these assumptions, it is especially challenging to improve on classes appearing later in the stream while remaining accurate on previous ones. This is due to the infamous problem of catastrophic forgetting, which causes a quick performance degradation when the classifier focuses on learning new categories. Recent literature proposed various approaches to tackle this issue, often resorting to very sophisticated techniques. In this work, we show that naive rehearsal can be patched to achieve similar performance. We point out some shortcomings that restrain Experience Replay (ER) and propose five tricks to mitigate them. Experiments show that ER, thus enhanced, displays an accuracy gain of 51.2 and 26.9 percentage points on the CIFAR-10 and CIFAR-100 datasets respectively (memory buffer size 1000). As a result, it surpasses current state-of-the-art rehearsal-based methods.

High-dimensional nonconvex problems are popular in today's machine learning and statistical genetics research. Recently, Ghadimi and Lan [1] proposed an algorithm to optimize nonconvex high-dimensional problems. There are several parameters in their algorithm that are to be set before running the algorithm. It is not trivial how to choose these parameters nor there is, to the best of our knowledge, an explicit rule on how to select the parameters to make the algorithm converges faster. We analyze Ghadimi and Lan's algorithm to gain an interpretation based on the inequality constraints for convergence and the upper bound for the norm of the gradient analogue. Our interpretation of their algorithm suggests this to be a damped Nesterov's acceleration scheme. Based on this, we propose an approach on how to select the parameters to improve convergence of the algorithm. Our numerical studies using high-dimensional nonconvex sparse learning problems, motivated by image denoising and statistical genetics applications, show that convergence can be made, on average, considerably faster than that of the conventional ISTA algorithm for such optimization problems with over $10000$ variables should the parameters be chosen using our proposed approach.

We aim to bridge the gap between typical human and machine-learning environments by extending the standard framework of few-shot learning to an online, continual setting. In this setting, episodes do not have separate training and testing phases, and instead models are evaluated online while learning novel classes. As in the real world, where the presence of spatiotemporal context helps us retrieve learned skills in the past, our online few-shot learning setting also features an underlying context that changes throughout time. Object classes are correlated within a context and inferring the correct context can lead to better performance. Building upon this setting, we propose a new few-shot learning dataset based on large scale indoor imagery that mimics the visual experience of an agent wandering within a world. Furthermore, we convert popular few-shot learning approaches into online versions and we also propose a new contextual prototypical memory model that can make use of spatiotemporal contextual information from the recent past.

In this paper, we propose a practical online method for solving a distributionally robust optimization (DRO) for deep learning, which has important applications in machine learning for improving the robustness of neural networks. In the literature, most methods for solving DRO are based on stochastic primal-dual methods. However, primal-dual methods for deep DRO suffer from several drawbacks: (1) manipulating a high-dimensional dual variable corresponding to the size of data is time expensive; (2) they are not friendly to online learning where data is coming sequentially. To address these issues, we transform the min-max formulation into a minimization formulation and propose a practical duality-free online stochastic method for solving deep DRO with KL divergence regularization. The proposed online stochastic method resembles the practical stochastic Nesterov's method in several perspectives that are widely used for learning deep neural networks. Under a Polyak-Lojasiewicz (PL) condition, we prove that the proposed method can enjoy an optimal sample complexity and a better round complexity (the number of gradient evaluations divided by a fixed mini-batch size) with a moderate mini-batch size than existing algorithms for solving the min-max or min formulation of DRO. Of independent interest, the proposed method can be also used for solving a family of stochastic compositional problems.

Detection of semantic similarity plays a vital role in sentence matching. It requires to learn discriminative representations of natural language. Recently, owing to more and more sophisticated model architecture, impressive progress has been made, along with a time-consuming training process and not-interpretable inference. To alleviate this problem, we explore a metric learning approach, named FILM (Fast, Interpretable, and Low-rank Metric learning) to efficiently find a high discriminative projection of the high-dimensional data. We construct this metric learning problem as a manifold optimization problem and solve it with the Cayley transformation method with the Barzilai-Borwein step size. In experiments, we apply FILM with triplet loss minimization objective to the Quora Challenge and Semantic Textual Similarity (STS) Task. The results demonstrate that the FILM method achieves superior performance as well as the fastest computation speed, which is consistent with our theoretical analysis of time complexity.

Sign language translation (SLT) aims to interpret sign video sequences into textbased natural language sentences. Sign videos consist of continuous sequences of sign gestures with no clear boundaries in between. Existing SLT models usually represent sign visual features in a frame-wise manner so as to avoid needing to explicitly segmenting the videos into isolated signs. However, these methods neglect the temporal information of signs and lead to substantial ambiguity in translation. In this paper, we explore the temporal semantic structures of sign videos to learn more discriminative features. To this end, we first present a novel sign video segment representation which takes into account multiple temporal granularities, thus alleviating the need for accurate video segmentation. Taking advantage of the proposed segment representation, we develop a novel hierarchical sign video feature learning method via a temporal semantic pyramid network, called TSPNet. Specifically, TSPNet introduces an inter-scale attention to evaluate and enhance local semantic consistency of sign segments and an intra-scale attention to resolve semantic ambiguity by using non-local video context. Experiments show that our TSPNet outperforms the state-of-the-art with significant improvements on the BLEU score (from 9.58 to 13.41) and ROUGE score (from 31.80 to 34.96) on the largest commonly-used SLT dataset.

Long training time hinders the potential of the deep Spiking Neural Network (SNN) with the online learning capability to be realized on the embedded systems hardware. Our work proposes a novel connection pruning approach that can be applied during the online Spike Timing Dependent Plasticity (STDP)-based learning to optimize the learning time and the network connectivity of the SNN. Our connection pruning approach was evaluated on a deep SNN with the Time To First Spike (TTFS) coding and has successfully achieved 2.1x speed-up in the online learning and reduced the network connectivity by 92.83%. The energy consumption in the online learning was saved by 64%. Moreover, the connectivity reduction results in 2.83x speed-up and 78.24% energy saved in the inference. Meanwhile, the classification accuracy remains the same as our non-pruning baseline on the Caltech 101 dataset. In addition, we developed an event-driven hardware architecture on the Field Programmable Gate Array (FPGA) platform that efficiently incorporates our proposed connection pruning approach while incurring as little as 0.56% power overhead. Moreover, we performed a comparison between our work and the existing works on connection pruning for SNN to highlight the key features of each approach. To the best of our knowledge, our work is the first to propose a connection pruning algorithm that can be applied during the online STDP-based learning for a deep SNN with the TTFS coding.