At a high level, the American AI Initiative seems to be headed in the right direction. We absolutely need a holistic approach that considers all the various areas that are critical to building innovative AI solutions. This seems to be an underlying concept of the Initiative, as the executive order places priority on making data available across government agencies, allocating cloud computing resources to support AI R&D and training the workforce. Commitment to AI innovation is critical to maintaining our leadership position in technology with the increasing level of global AI competition. We know that China, France and the U.K. have invested and committed billions already to their own AI initiatives.

It is intuitive that semantic representations can be useful for machine translation, mainly because they can help in enforcing meaning preservation and handling data sparsity (many sentences correspond to one meaning) of machine translation models. On the other hand, little work has been done on leveraging semantics for neural machine translation (NMT). In this work, we study the usefulness of AMR (short for abstract meaning representation) on NMT. Experiments on a standard English-to-German dataset show that incorporating AMR as additional knowledge can significantly improve a strong attention-based sequence-to-sequence neural translation model.

The performance of automatic speaker recognition systems degrades when facing distorted speech data containing additive noise and/or reverberation. Statistical uncertainty propagation has been introduced as a promising paradigm to address this challenge. So far, different uncertainty propagation methods have been proposed to compensate noise and reverberation in i-vectors in the context of speaker recognition. They have achieved promising results on small datasets such as YOHO and Wall Street Journal, but little or no improvement on the larger, highly variable NIST Speaker Recognition Evaluation (SRE) corpus. In this paper, we propose a complete uncertainty propagation method, whereby we model the effect of uncertainty both in the computation of unbiased Baum-Welch statistics and in the derivation of the posterior expectation of the i-vector. We conduct experiments on the NIST-SRE corpus mixed with real domestic noise and reverberation from the CHiME-2 corpus and preprocessed by multichannel speech enhancement. The proposed method improves the equal error rate (EER) by 4% relative compared to a conventional i-vector based speaker verification baseline. This is to be compared with previous methods which degrade performance.

Many machine learning algorithms represent input data with vector embeddings or discrete codes. When inputs exhibit compositional structure (e.g. objects built from parts or procedures from subroutines), it is natural to ask whether this compositional structure is reflected in the the inputs' learned representations. While the assessment of compositionality in languages has received significant attention in linguistics and adjacent fields, the machine learning literature lacks general-purpose tools for producing graded measurements of compositional structure in more general (e.g. vector-valued) representation spaces. We describe a procedure for evaluating compositionality by measuring how well the true representation-producing model can be approximated by a model that explicitly composes a collection of inferred representational primitives. We use the procedure to provide formal and empirical characterizations of compositional structure in a variety of settings, exploring the relationship between compositionality and learning dynamics, human judgments, representational similarity, and generalization.

To be effective in sequential data processing, Recurrent Neural Networks (RNNs) are required to keep track of past events by creating memories. While the relation between memories and the network's hidden state dynamics was established over the last decade, previous works in this direction were of a predominantly descriptive nature focusing mainly on locating the dynamical objects of interest. In particular, it remained unclear how dynamical observables affect the performance, how they form and whether they can be manipulated. Here, we utilize different training protocols, datasets and architectures to obtain a range of networks solving a delayed classification task with similar performance, alongside substantial differences in their ability to extrapolate for longer delays. We analyze the dynamics of the network's hidden state, and uncover the reasons for this difference. Each memory is found to be associated with a nearly steady state of the dynamics which we refer to as a 'slow point'. Slow point speeds predict extrapolation performance across all datasets, protocols and architectures tested. Furthermore, by tracking the formation of the slow points we are able to understand the origin of differences between training protocols. Finally, we propose a novel regularization technique that is based on the relation between hidden state speeds and memory longevity. Our technique manipulates these speeds, thereby leading to a dramatic improvement in memory robustness over time, and could pave the way for a new class of regularization methods.

Large-scale industrial recommender systems are usually confronted with computational problems due to the enormous corpus size. To retrieve and recommend the most relevant items to users under response time limits, resorting to an efficient index structure is an effective and practical solution. Tree-based Deep Model (TDM) for recommendation \cite{zhu2018learning} greatly improves recommendation accuracy using tree index. By indexing items in a tree hierarchy and training a user-node preference prediction model satisfying a max-heap like property in the tree, TDM provides logarithmic computational complexity w.r.t. the corpus size, enabling the use of arbitrary advanced models in candidate retrieval and recommendation. In tree-based recommendation methods, the quality of both the tree index and the trained user preference prediction model determines the recommendation accuracy for the most part. We argue that the learning of tree index and user preference model has interdependence. Our purpose, in this paper, is to develop a method to jointly learn the index structure and user preference prediction model. In our proposed joint optimization framework, the learning of index and user preference prediction model are carried out under a unified performance measure. Besides, we come up with a novel hierarchical user preference representation utilizing the tree index hierarchy. Experimental evaluations with two large-scale real-world datasets show that the proposed method improves recommendation accuracy significantly. Online A/B test results at Taobao display advertising also demonstrate the effectiveness of the proposed method in production environments.

In the past decade, sparse principal component analysis has emerged as an archetypal problem for illustrating statistical-computational tradeoffs. This trend has largely been driven by a line of research aiming to characterize the average-case complexity of sparse PCA through reductions from the planted clique (PC) conjecture - which conjectures that there is no polynomial-time algorithm to detect a planted clique of size $K = o(N^{1/2})$ in $\mathcal{G}(N, \frac{1}{2})$. All previous reductions to sparse PCA either fail to show tight computational lower bounds matching existing algorithms or show lower bounds for formulations of sparse PCA other than its canonical generative model, the spiked covariance model. Also, these lower bounds all quickly degrade with the exponent in the PC conjecture. Specifically, when only given the PC conjecture up to $K = o(N^\alpha)$ where $\alpha < 1/2$, there is no sparsity level $k$ at which these lower bounds remain tight. If $\alpha \le 1/3$ these reductions fail to even show the existence of a statistical-computational tradeoff at any sparsity $k$. We give a reduction from PC that yields the first full characterization of the computational barrier in the spiked covariance model, providing tight lower bounds at all sparsities $k$. We also show the surprising result that weaker forms of the PC conjecture up to clique size $K = o(N^\alpha)$ for any given $\alpha \in (0, 1/2]$ imply tight computational lower bounds for sparse PCA at sparsities $k = o(n^{\alpha/3})$. This shows that even a mild improvement in the signal strength needed by the best known polynomial-time sparse PCA algorithms would imply that the hardness threshold for PC is subpolynomial. This is the first instance of a suboptimal hardness assumption implying optimal lower bounds for another problem in unsupervised learning.

While deep networks have strong fitting capability to complex input patterns, they can easily overfit tobiased training data with corrupted labels. Sample reweighting strategy is commonly used to alleviate this robust learning issue, through imposing zeroor possibly smaller weights to corrupted samples to suppress their negative influence to learning. Current reweighting algorithms, however, needelaborate tuning of additional hyperparameters orcareful designing of complex metalearner for learning to assign weights on samples. To address these issues, we propose a new metalearning methodwith few tuned hyper-parameters and simple structure of a meta-learner (one hidden layer MLP network). Guided by a small amount of unbiased metadata, the parameters of the proposed meta-learnercan be gradually evolved for finely tugging the classifier gradient approaching tothe right direction. This learning manner complies with a real teaching progress: A good teacher should more respect the student's own learning manner and help progressively correct his learning bias based on his/her current learning status.Experimental results substantiate the robustness of the new algorithm on corrupted label cases,as well as its stability and efficiency in learning. The dream begins with a teacher who believes in you, who tugs and pushes and leads you to the next plateau, sometimes pokingyou with a sharp stick called truth.

Traditional load analysis is facing challenges with the new electricity usage patterns due to demand response as well as increasing deployment of distributed generations, including photovoltaics (PV), electric vehicles (EV), and energy storage systems (ESS). At the transmission system, despite of irregular load behaviors at different areas, highly aggregated load shapes still share similar characteristics. Load clustering is to discover such intrinsic patterns and provide useful information to other load applications, such as load forecasting and load modeling. This paper proposes an efficient submodular load clustering method for transmission-level load areas. Robust principal component analysis (R-PCA) firstly decomposes the annual load profiles into low-rank components and sparse components to extract key features. A novel submodular cluster center selection technique is then applied to determine the optimal cluster centers through constructed similarity graph. Following the selection results, load areas are efficiently assigned to different clusters for further load analysis and applications. Numerical results obtained from PJM load demonstrate the effectiveness of the proposed approach.

Thompson sampling (TS) is a class of algorithms for sequential decision-making, which requires maintaining a posterior distribution over a model. However, calculating exact posterior distributions is intractable for all but the simplest models. Consequently, efficient computation of an approximate posterior distribution is a crucial problem for scalable TS with complex models, such as neural networks. In this paper, we use distribution optimization techniques to approximate the posterior distribution, solved via Wasserstein gradient flows. Based on the framework, a principled particle-optimization algorithm is developed for TS to approximate the posterior efficiently. Our approach is scalable and does not make explicit distribution assumptions on posterior approximations. Extensive experiments on both synthetic data and real large-scale data demonstrate the superior performance of the proposed methods.