Search

Deep generative models (DGM) combine the deep neural networks with generative models to learn the underlying generation mechanism of the data of interest. This has emerged as a important approach to extracting knowledge from data in machine learning and artificial intelligence. However, there are many challenges to learning and applying DGMs in different domains despite their promising potential. Therefore, this thesis focuses on understanding, improving, and applying different deep generative models. First, we introduce the underlying principles under different DGMs, including the Variational Auto-Encoder (VAE), the Flow-Based Model, the Generative Adversarial Network (GAN), and the Energy-Based Model (EBM). We also propose a novel counterpart of VAEs: Variational Latent Optimization (VLO), which does not require an encoder structure. Besides, we provide a new angle to understand the generation process of EBMs, build a connection between EBMs and GANs, and design a new approach to improve the sample quality of EBMs from that. Next, we propose two hybrids of DGMs to improve the generation quality of current models. First, we combine Flow-based models and Variational Auto-Encoders to improve the generation quality of Auto-Encoder-based generative models. Second, we borrow the idea of exponential tilting and combine the energy-based model with other likelihood-based generative models to obtain better samples. In the end, we conduct various applications related to modern deep generative models, including using generative models as likelihood-based methods for out-of-distribution (OOD) detection and designing controllable generative models over human faces. We propose a novel OOD-detect score called likelihood regret to help detect OOD samples with VAEs. Also, we propose to add a new structure to current key-points-based face reenactment models and combine them with the 3D morphable model to improve their generation quality and generalization ability.

This dissertation investigates the performance of hedge funds and actively managed U.S. equity mutual funds. The first chapter examines the relation between hedge funds and the low beta anomaly. Different conditions in the mutual fund and hedge fund industries should lead to different approaches with respect to the low beta anomaly. I find that, unlike most mutual funds, the average hedge fund tends to benefit considerably from the anomaly. About 2.3% per year of apparent alpha for the average hedge fund can be attributed to the low beta anomaly rather than manager skill. Low skill managers are the most reliant on the anomaly to generate returns, with the most reliant underperforming the least reliant by 5.9% per year. The second chapter uses machine learning to dynamically identify and optimally combine the predictors of hedge fund performance. The portfolio formed based on the machine learning models has an out-of-sample alpha of 7.8% per year. The importance of each predictor varies over time, but among the 22 predictors I consider, the consistently important predictors are average return, maximum return, alpha, systematic risk, and beta activity. Machine learning provides valuable, unique information about future hedge fund performance that is not captured by individual predictors. The third chapter studies whether the quality of fund risk management can predict fund performance. I find that the risk management skills of mutual fund managers—as quantified by their funds’ maximum drawdowns—are persistent and predictive of subsequent risk-adjusted performance. Funds with relatively strong past performance and relatively low past maximum drawdowns have, on average, an out-of-sample alpha of 2.68% per year. That alpha is magnified when markets are turbulent—a time during which risk management skills should be most valuable. Investors are averse to drawdown risk. After controlling for typical measures of past performance, fund flows are still a decreasing function of maximum drawdowns, particularly among investors with greater risk aversion and during times of generally heightened risk aversion.