Database performance troubleshooting is a complex multi-step process that broadly involves three key stages- (a) Detection: determining what's wrong and when; (b) Root Cause Analysis (RCA): reasoning about why is the performance poor; (c) Resolution: identifying a fix. A plethora of techniques exist to address each of these problems, but they hardly work in real-world at scale. First, real-world customer workloads are noisy, non-stationary and quasi-periodic in nature rendering traditional detectors ineffective. Second, real-world production databases execute a highly diverse set of queries that skew the database statistics into long-tail distributions causing traditional RCA methods to fail. Third, these databases typically execute millions of such diverse queries every minute rendering traditional methods inefficient when deployed at scale.

Technology improvements have made it easier than ever to collect diverse telemetry at high resolution from any cyber or physical system, for both monitoring and control. In the domain of monitoring, anomaly detection has become an important problem in many research areas ranging from IoT and sensor networks to devOps. These systems operate in real, noisy and non-stationary environments. A fundamental question is then, ‘How to quickly spot anomalies in a data-stream, and differentiate them from either sudden or gradual drifts in the normal behaviour?’ Although several heuristics have been proposed for detecting anomalies on streams, no known method has formalized the desiderata and rigorously proven that they can be achieved. We begin by formalizing the problem as a sequential estimation task. We propose \name, (\textbf{Fi}ne \textbf{T}une on \textbf{Ne}w and \textbf{S}imilar \textbf{S}amples), a flexible framework for detecting anomalies on data streams. We show that in the case when the data stream has a gaussian distribution, FITNESS is provably both robust and adaptive. The core of our method is to fine-tune the anomaly detection system only on recent, similar examples, before predicting an anomaly score. We prove that this is sufficient for robustness and adaptivity. We further experimentally demonstrate that \name;{is} flexible in practice, i.e., it can convert existing offline AD algorithms in to robust and adaptive online ones.

In this work, we propose ParaNet, a non-autoregressive seq2seq model that converts text to spectrogram. It is fully convolutional and brings 46.7 times speed-up over the lightweight Deep Voice 3 at synthesis, while obtaining reasonably good speech quality. ParaNet also produces stable alignment between text and speech on the challenging test sentences by iteratively improving the attention in a layer-by-layer manner. Furthermore, we build the parallel text-to-speech system by applying various parallel neural vocoders, which can synthesize speech from text through a single feed-forward pass. We also explore a novel VAE-based approach to train the inverse autoregressive flow (IAF) based parallel vocoder from scratch, which avoids the need for distillation from a separately trained WaveNet as previous work.

In this work, we propose WaveFlow, a small-footprint generative flow for raw audio, which is directly trained with maximum likelihood. It handles the long-range structure of 1-D waveform with a dilated 2-D convolutional architecture, while modeling the local variations using expressive autoregressive functions. WaveFlow provides a unified view of likelihood-based models for 1-D data, including WaveNet and WaveGlow as special cases. It generates high-fidelity speech as WaveNet, while synthesizing several orders of magnitude faster as it only requires a few sequential steps to generate very long waveforms with hundreds of thousands of time-steps. Furthermore, it can significantly reduce the likelihood gap that has existed between autoregressive models and flow-based models for efficient synthesis. Finally, our small-footprint WaveFlow has only 5.91M parameters, which is 15{\texttimes} smaller than WaveGlow. It can generate 22.05 kHz high-fidelity audio 42.6{\texttimes} faster than real-time (at a rate of 939.3 kHz) on a V100 GPU without engineered inference kernels.

The impact of softmax on the value function itself in reinforcement learning (RL) is often viewed as problematic because it leads to sub-optimal value (or Q) functions and interferes with the contraction properties of the Bellman operator. Surprisingly, despite these concerns, and independent of its effect on exploration, the softmax Bellman operator when combined with Deep Q-learning, leads to Q-functions with superior policies in practice, even outperforming its double Q-learning counterpart. To better understand how and why this occurs, we revisit theoretical properties of the softmax Bellman operator, and prove that (i) it converges to the standard Bellman operator exponentially fast in the inverse temperature parameter, and (ii) the distance of its Q function from the optimal one can be bounded. These alone do not explain its superior performance, so we also show that the softmax operator can reduce the overestimation error, which may give some insight into why a sub-optimal operator leads to better performance in the presence of value function approximation. A comparison among different Bellman operators is then presented, showing the trade-offs when selecting them.

We propose a scalable algorithm for model selection in sigmoid belief networks (SBNs), based on the factorized asymptotic Bayesian (FAB) framework. We derive the corresponding generalized factorized information criterion (gFIC) for the SBN, which is proven to be statistically consistent with the marginal log-likelihood. To capture the dependencies within hidden variables in SBNs, a recognition network is employed to model the variational distribution. The resulting algorithm, which we call FABIA, can simultaneously execute both model selection and inference by maximizing the lower bound of gFIC. On both synthetic and real data, our experiments suggest that FABIA, when compared to state-of-the-art algorithms for learning SBNs, $(i)$ produces a more concise model, thus enabling faster testing; $(ii)$ improves predictive performance; $(iii)$ accelerates convergence; and $(iv)$ prevents overfitting.

Belief networks are commonly used generative models of data, but require expensive posterior estimation to train and test the model. Learning typically proceeds by posterior sampling, variational approximations, or recognition networks, combined with stochastic optimization. We propose using an online Monte Carlo expectation-maximization (MCEM) algorithm to learn the maximum a posteriori (MAP) estimator of the generative model or optimize the variational lower bound of a recognition network. The E-step in this algorithm requires posterior samples, which are already generated in current learning schema. For the M-step, we augment with Polya-Gamma (PG) random variables to give an analytic updating scheme. We show relationships to standard learning approaches by deriving stochastic gradient ascent in the MCEM framework. We apply the proposed methods to both binary and count data. Experimental results show that MCEM improves the convergence speed and often improves hold-out performance over existing learning methods. Our approach is readily generalized to other recognition networks.

Feature construction is of vital importance in reinforcement learning, as the quality of a value function or policy is largely determined by the corresponding features. The recent successes of deep reinforcement learning (RL) only increase the importance of understanding feature construction. Typical deep RL approaches use a linear output layer, which means that deep RL can be interpreted as a feature construction/encoding network followed by linear value function approximation. This paper develops and evaluates a theory of linear feature encoding. We extend theoretical results on feature quality for linear value function approximation from the uncontrolled case to the controlled case. We then develop a supervised linear feature encoding method that is motivated by insights from linear value function approximation theory, as well as empirical successes from deep RL. The resulting encoder is a surprisingly effective method for linear value function approximation using raw images as inputs.