T raditionally, mathematical formulations of dynamical systems in the context of Signal Processing and Control Theory have been a lynchpin of today's Financial Engineering. More recently, advances in sequential decision making, mainly through the concept of Reinforcement Learning, have been instrumental in the development of multistage stochastic optimization, a key component in sequential portfolio optimization (asset allocation) strategies. In this thesis, we develop a comprehensive account of the expressive power, modelling efficiency, and performance advantages of so called trading agents (i.e., Deep Soft Recurrent Q-Network (DSRQN) and Mixture of Score Machines (MSM)), based on both traditional system identification (model-based approach) as well as on context-independent agents (model-free approach). The analysis provides a conclusive support for the ability of model-free reinforcement learning methods to act as universal trading agents, which are not only capable of reducing the computational and memory complexity (owing to their linear scaling with size of the universe), but also serve as generalizing strategies across assets and markets, regardless of the trading universe on which they have been trained. The relatively low volume of daily returns in financial market data is addressed via data augmentation (a generative approach) and a choice of pre-training strategies, both of which are validated against current state-of-the-art models. For rigour, a risk-sensitive framework which includes transaction costs is considered, and its performance advantages are demonstrated in a variety of scenarios, from synthetic time-series (sinusoidal, sawtooth and chirp waves), ii simulated market series (surrogate data based), through to real market data (S&P 500 and EURO STOXX 50). The analysis and simulations confirm the superiority of universal model-free reinforcement learning agents over current portfolio management model in asset allocation strategies, with the achieved performance advantage of as much as 9.2% in annualized cumulative returns and 13.4% in annualized Sharpe Ratio.

Recommender systems are personalized: we expect the results given to a particular user to reflect that user's preferences. Some researchers have studied the notion of calibration, how well recommendations match users' stated preferences, and bias disparity the extent to which mis-calibration affects different user groups. In this paper, we examine bias disparity over a range of different algorithms and for different item categories and demonstrate significant differences between model-based and memory-based algorithms.

We introduce an approach for imposing physically motivated inductive biases on graph networks to learn interpretable representations and improved zero-shot generalization. Our experiments show that our graph network models, which implement this inductive bias, can learn message representations equivalent to the true force vector when trained on n-body gravitational and spring-like simulations. We use symbolic regression to fit explicit algebraic equations to our trained model's message function and recover the symbolic form of Newton's law of gravitation without prior knowledge. We also show that our model generalizes better at inference time to systems with more bodies than had been experienced during training. Our approach is extensible, in principle, to any unknown interaction law learned by a graph network, and offers a valuable technique for interpreting and inferring explicit causal theories about the world from implicit knowledge captured by deep learning.

Effective and powerful methods for denoising real electrocardiogram (ECG) signals are important for wearable sensors and devices. Deep Learning (DL) models have been used extensively in image processing and other domains with great success but only very recently have been used in processing ECG signals. This paper presents several DL models namely Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM), Restricted Boltzmann Machine (RBM) together with the more conventional filtering methods (low pass filtering, high pass filtering, Notch filtering) and the standard wavelet-based technique for denoising EEG signals. These methods are trained, tested and evaluated on different synthetic and real ECG datasets taken from the MIT PhysioNet database and for different simulation conditions (i.e. various lengths of the ECG signals, single or multiple records). The results show the CNN model is a performant model that can be used for off-line denoising ECG applications where it is satisfactory to train on a clean part of an ECG signal from an ECG record, and then to test on the same ECG signal, which would have some high level of noise added to it. However, for real-time applications or near-real time applications, this task becomes more cumbersome, as the clean part of an ECG signal is very probable to be very limited in size. Therefore the solution put forth in this work is to train a CNN model on 1 second ECG noisy artificial multiple heartbeat data (i.e. ECG at effort), which was generated in a first instance based on few sequences of real signal heartbeat ECG data (i.e. ECG at rest). Afterwards it would be possible to use the trained CNN model in real life situations to denoise the ECG signal.

Unsupervised recalibration (URC) is a general way to improve the accuracy of an already trained probabilistic classification or regression model upon encountering new data while deployed in the field. URC does not require any ground truth associated with the new field data. URC merely observes the model's predictions and recognizes when the training set is not representative of field data, and then corrects to remove any introduced bias. URC can be particularly useful when applied separately to different subpopulations observed in the field that were not considered as features when training the machine learning model. This makes it possible to exploit subpopulation information without retraining the model or even having ground truth for some or all subpopulations available. Additionally, if these subpopulations are the object of study, URC serves to determine the correct ground truth distributions for them, where naive aggregation methods, like averaging the model's predictions, systematically underestimate their differences.

Each block contains a certain amount of data; and the chain connects these data to form a distributed database. New block needs to be approved by all or some of the network members to become a valid block. Each approved block includes information of the previous block in the chain, therefore if the block is changed, all blocks before this block will be invalid as well. The strategies to reach agreement of the new block (consensus) vary in different types of blockchain. The mathematical structure of the blockchain implies two essential properties: (i) the data (in block) is immutable; (ii) the distributed network with consensus allows users to communicate directly with each other and download a copy of the current ledger, which means that there is continuous monitoring and redundancy of the data in the network.

Enabling robots to walk and run on yielding terrain is increasingly vital to endeavors ranging from disaster response to extraterrestrial exploration. While dynamic legged locomotion on rigid ground is challenging enough, yielding terrain presents additional challenges such as permanent ground deformation which dissipates energy. In this paper, we examine the soft landing problem: given some impact momentum, bring the robot to rest while minimizing foot penetration depth. To gain insight into properties of penetration depth-minimizing control policies, we formulate a constrained optimal control problem and obtain a bang-bang open-loop force profile. Motivated by examples from biology and recent advances in legged robotics, we also examine impedance-control solutions to the dimensionless soft landing problem. Through simulations, we find that optimal impedance reduces penetration depth nearly as much as the open-loop force profile, while remaining robust to model uncertainty. Through simulations and experiments, we find that the solution space is rich, exhibiting qualitatively different relationships between impact velocity and the optimal impedance for small and large dimensionless impact velocities. Lastly, we discuss the relevance of this work to minimum-cost-of-transport locomotion for several actuator design choices.

Monitoring patients in ICU is a challenging and high-cost task. Hence, predicting the condition of patients during their ICU stay can help provide better acute care and plan the hospital's resources. There has been continuous progress in machine learning research for ICU management, and most of this work has focused on using time series signals recorded by ICU instruments. In our work, we show that adding clinical notes as another modality improves the performance of the model for three benchmark tasks: in-hospital mortality prediction, modeling decompensation, and length of stay forecasting that play an important role in ICU management. While the time-series data is measured at regular intervals, doctor notes are charted at irregular times, making it challenging to model them together. We propose a method to model them jointly, achieving considerable improvement across benchmark tasks over baseline time-series model. Our implementation can be found at \url{https://github.com/kaggarwal/ClinicalNotesICU}.

Question semantic similarity (Q2Q) is a challenging task that is very useful in many NLP applications, such as detecting duplicate questions and question answering systems. In this paper, we present the results and findings of the shared task (Semantic Question Similarity in Arabic). The task was organized as part of the first workshop on NLP Solutions for Under Resourced Languages (NSURL 2019) The goal of the task is to predict whether two questions are semantically similar or not, even if they are phrased differently. A total of 9 teams participated in the task. The datasets created for this task are made publicly available to support further research on Arabic Q2Q.

To improve the execution speed and efficiency of neural networks in embedded systems, it is crucial to decrease the model size and computational complexity. In addition to conventional compression techniques, e.g., weight pruning and quantization, removing unimportant activations can reduce the amount of data communication and the computation cost. Unlike weight parameters, the pattern of activations is directly related to input data and thereby changes dynamically. To regulate the dynamic activation sparsity (DAS), in this work, we propose a generic low-cost approach based on winners-take-all (WTA) dropout technique. The network enhanced by the proposed WTA dropout, namely \textit{DASNet}, features structured activation sparsity with an improved sparsity level. Compared to the static feature map pruning methods, DASNets provide better computation cost reduction. The WTA technique can be easily applied in deep neural networks without incurring additional training variables. More importantly, DASNet can be seamlessly integrated with other compression techniques, such as weight pruning and quantization, without compromising on accuracy. Our experiments on various networks and datasets present significant run-time speedups with negligible accuracy loss.