Intelligent Tutoring Systems (ITSs) have been developed to provide students with personalized learning experiences by adaptively generating learning paths optimized for each individual. Within the vast scope of ITS, score prediction stands out as an area of study that enables students to construct individually realistic goals based on their current position. Via the expected score provided by the ITS, a student can instantaneously compare one's expected score to one's actual score, which directly corresponds to the reliability that the ITS can instill. In other words, refining the precision of predicted scores strictly correlates to the level of confidence that a student may have with an ITS, which will evidently ensue improved student engagement. However, previous studies have solely concentrated on improving the performance of a prediction model, largely lacking focus on the benefits generated by its practical application. In this paper, we demonstrate that the accuracy of the score prediction model deployed in a real-world setting significantly impacts user engagement by providing empirical evidence. To that end, we apply a state-of-the-art deep attentive neural network-based score prediction model to Santa, a multi-platform English ITS with approximately 780K users in South Korea that exclusively focuses on the TOEIC (Test of English for International Communications) standardized examinations. We run a controlled A/B test on the ITS with two models, respectively based on collaborative filtering and deep attentive neural networks, to verify whether the more accurate model engenders any student engagement. The results conclude that the attentive model not only induces high student morale (e.g. higher diagnostic test completion ratio, number of questions answered, etc.) but also encourages active engagement (e.g. higher purchase rate, improved total profit, etc.) on Santa.

Massive language models are the core of modern NLP modeling and have been shown to encode impressive amounts of commonsense and factual information. However, that knowledge exists only within the latent parameters of the model, inaccessible to inspection and interpretation, and even worse, factual information memorized from the training corpora is likely to become stale as the world changes. Knowledge stored as parameters will also inevitably exhibit all of the biases inherent in the source materials. To address these problems, we develop a neural language model that includes an explicit interface between symbolically interpretable factual information and subsymbolic neural knowledge. We show that this model dramatically improves performance on two knowledge-intensive question-answering tasks. More interestingly, the model can be updated without re-training by manipulating its symbolic representations. In particular this model allows us to add new facts and overwrite existing ones in ways that are not possible for earlier models.

Research in developmental psychology consistently shows that children explore the world thoroughly and efficiently and that this exploration allows them to learn. In turn, this early learning supports more robust generalization and intelligent behavior later in life. While much work has gone into developing methods for exploration in machine learning, artificial agents have not yet reached the high standard set by their human counterparts. In this work we propose using DeepMind Lab (Beattie et al., 2016) as a platform to directly compare child and agent behaviors and to develop new exploration techniques. We outline two ongoing experiments to demonstrate the effectiveness of a direct comparison, and outline a number of open research questions that we believe can be tested using this methodology.

The existence of adversarial examples capable of fooling trained neural network classifiers calls for a much better understanding of possible attacks, in order to guide the development of safeguards against them. It includes attack methods in the highly challenging non-interactive blackbox setting, where adversarial attacks are generated without any access, including queries, to the target model. Prior works in this setting have relied mainly on algorithmic innovations derived from empirical observations (e.g., that momentum helps), and the field currently lacks a firm theoretical basis for understanding transferability in adversarial attacks. In this work, we address this gap and lay the theoretical foundations for crafting transferable adversarial examples to entire function classes. We introduce Adversarial Examples Games (AEG), a novel framework that models adversarial examples as two-player min-max games between an attack generator and a representative classifier. We prove that the saddle point of an AEG game corresponds to a generating distribution of adversarial examples against entire function classes. Training the generator only requires the ability to optimize a representative classifier from a given hypothesis class, enabling BlackBox transfer to unseen classifiers from the same class. We demonstrate the efficacy of our approach on the MNIST and CIFAR-10 datasets against both undefended and robustified models, achieving competitive performance with state-of-the-art BlackBox transfer approaches.

We examine Dropout through the perspective of interactions: learned effects that combine multiple input variables. Given $N$ variables, there are $O(N^2)$ possible pairwise interactions, $O(N^3)$ possible 3-way interactions, etc. We show that Dropout implicitly sets a learning rate for interaction effects that decays exponentially with the size of the interaction, corresponding to a regularizer that balances against the hypothesis space which grows exponentially with number of variables in the interaction. This understanding of Dropout has implications for the optimal Dropout rate: higher Dropout rates should be used when we need stronger regularization against spurious high-order interactions. This perspective also issues caution against using Dropout to measure term saliency because Dropout regularizes against terms for high-order interactions. Finally, this view of Dropout as a regularizer of interaction effects provides insight into the varying effectiveness of Dropout for different architectures and data sets. We also compare Dropout to regularization via weight decay and early stopping and find that it is difficult to obtain the same regularization effect for high-order interactions with these methods.

Accurately forecasting county level COVID-19 confirmed cases is crucial to optimizing medical resources. Forecasting emerging outbreaks pose a particular challenge because many existing forecasting techniques learn from historical seasons trends. Recurrent neural networks (RNNs) with LSTM-based cells are a logical choice of model due to their ability to learn temporal dynamics. In this paper, we adapt the state and county level influenza model, TDEFSI-LONLY, proposed in Wang et a. [l2020] to national and county level COVID-19 data. We show that this model poorly forecasts the current pandemic. We analyze the two week ahead forecasting capabilities of the TDEFSI-LONLY model with combinations of regularization techniques. Effective training of the TDEFSI-LONLY model requires data augmentation, to overcome this challenge we utilize an SEIR model and present an inter-county mixing extension to this model to simulate sufficient training data. Further, we propose an alternate forecast model, {\it County Level Epidemiological Inference Recurrent Network} (\alg{}) that trains an LSTM backbone on national confirmed cases to learn a low dimensional time pattern and utilizes a time distributed dense layer to learn individual county confirmed case changes each day for a two weeks forecast. We show that the best, worst, and median state forecasts made using CLEIR-Net model are respectively New York, South Carolina, and Montana.

Accurate and reliable prediction of hospital admission location is important due to resource-constraints and space availability in a clinical setting, particularly when dealing with patients who come from the emergency department. In this work we propose a student-teacher network via reinforcement learning to deal with this specific problem. A representation of the weights of the student network is treated as the state and is fed as an input to the teacher network. The teacher network's action is to select the most appropriate batch of data to train the student network on from a training set sorted according to entropy. By validating on three datasets, not only do we show that our approach outperforms state-of-the-art methods on tabular data and performs competitively on image recognition, but also that novel curricula are learned by the teacher network. We demonstrate experimentally that the teacher network can actively learn about the student network and guide it to achieve better performance than if trained alone.

Deep neural networks are typically trained under a supervised learning framework where a model learns a single task using labeled data. Instead of relying solely on labeled data, practitioners can harness unlabeled or related data to improve model performance, which is often more accessible and ubiquitous. Self-supervised pre-training for transfer learning is becoming an increasingly popular technique to improve state-of-the-art results using unlabeled data. It involves first pre-training a model on a large amount of unlabeled data, then adapting the model to target tasks of interest. In this review, we survey self-supervised learning methods and their applications within the sequential transfer learning framework. We provide an overview of the taxonomy for self-supervised learning and transfer learning, and highlight some prominent methods for designing pre-training tasks across different domains. Finally, we discuss recent trends and suggest areas for future investigation.

Training neural networks with many processors can reduce time-to-solution; however, it is challenging to maintain convergence and efficiency at large scales. The Kronecker-factored Approximate Curvature (K-FAC) was recently proposed as an approximation of the Fisher Information Matrix that can be used in natural gradient optimizers. We investigate here a scalable K-FAC design and its applicability in convolutional neural network (CNN) training at scale. We study optimization techniques such as layer-wise distribution strategies, inverse-free second-order gradient evaluation, and dynamic K-FAC update decoupling to reduce training time while preserving convergence. We use residual neural networks (ResNet) applied to the CIFAR-10 and ImageNet-1k datasets to evaluate the correctness and scalability of our K-FAC gradient preconditioner. With ResNet-50 on the ImageNet-1k dataset, our distributed K-FAC implementation converges to the 75.9% MLPerf baseline in 18-25% less time than does the classic stochastic gradient descent (SGD) optimizer across scales on a GPU cluster.

Small integration time steps limit molecular dynamics (MD) simulations to millisecond time scales. Markov state models (MSMs) and equation-free approaches learn low-dimensional kinetic models from MD simulation data by performing configurational or dynamical coarse-graining of the state space. The learned kinetic models enable the efficient generation of dynamical trajectories over vastly longer time scales than are accessible by MD, but the discretization of configurational space and/or absence of a means to reconstruct molecular configurations precludes the generation of continuous all-atom molecular trajectories. We propose latent space simulators (LSS) to learn kinetic models for continuous all-atom simulation trajectories by training three deep learning networks to (i) learn the slow collective variables of the molecular system, (ii) propagate the system dynamics within this slow latent space, and (iii) generatively reconstruct molecular configurations. We demonstrate the approach in an application to Trp-cage miniprotein to produce novel ultra-long synthetic folding trajectories that accurately reproduce all-atom molecular structure, thermodynamics, and kinetics at six orders of magnitude lower cost than MD. The dramatically lower cost of trajectory generation enables greatly improved sampling and greatly reduced statistical uncertainties in estimated thermodynamic averages and kinetic rates.