In this paper, we study feature cross search as a fundamental primitive in feature engineering. The importance of feature cross search especially for the linear model has been known for a while, with well-known textbook examples. In this problem, the goal is to select a small subset of features, combine them to form a new feature (called the crossed feature) by considering their Cartesian product, and find feature crosses to learn an \emph{accurate} model. In particular, we study the problem of maximizing a normalized Area Under the Curve (AUC) of the linear model trained on the crossed feature column. First, we show that it is not possible to provide an $n^{1/\log\log n}$-approximation algorithm for this problem unless the exponential time hypothesis fails. This result also rules out the possibility of solving this problem in polynomial time unless $\mathsf{P}=\mathsf{NP}$. On the positive side, by assuming the \naive\ assumption, we show that there exists a simple greedy $(1-1/e)$-approximation algorithm for this problem. This result is established by relating the AUC to the total variation of the commutator of two probability measures and showing that the total variation of the commutator is monotone and submodular. To show this, we relate the submodularity of this function to the positive semi-definiteness of a corresponding kernel matrix. Then, we use Bochner's theorem to prove the positive semi-definiteness by showing that its inverse Fourier transform is non-negative everywhere. Our techniques and structural results might be of independent interest.

In his book'The Master Algorithm', Pedro Domingos dissects the ML trains of thought into 5 tribes/groups. Symbolists believe in inductive logic, using data and rules to model intelligent systems. This can be computationally intensive at times, subjected to over-fitting and the'Bias-Variance' paradigm. However, experimentation is the key in most of the methods employed here, for e.g. Connectionists are interested in learning how the brain works and mimic the same using neural networks. Deep learning and mapping the brain (Neurosciences) is endogenous to this tribe.

Spectral embedding of network adjacency matrices often produces node representations living approximately around low-dimensional submanifold structures. In particular, hidden substructure is expected to arise when the graph is generated from a latent position model. Furthermore, the presence of communities within the network might generate community-specific submanifold structures in the embedding, but this is not explicitly accounted for in most statistical models for networks. In this article, a class of models called latent structure block models (LSBM) is proposed to address such scenarios, allowing for graph clustering when community-specific one dimensional manifold structure is present. LSBMs focus on a specific class of latent space model, the random dot product graph (RDPG), and assign a latent submanifold to the latent positions of each community. A Bayesian model for the embeddings arising from LSBMs is discussed, and shown to have a good performance on simulated and real world network data. The model is able to correctly recover the underlying communities living in a one-dimensional manifold, even when the parametric form of the underlying curves is unknown, achieving remarkable results on a variety of real data.

We establish a novel framework for learning a directed acyclic graph (DAG) when data are generated from a Gaussian, linear structural equation model. It consists of two parts: (1) introduce a permutation matrix as a new parameter within a regularized Gaussian log-likelihood to represent variable ordering; and (2) given the ordering, estimate the DAG structure through sparse Cholesky factor of the inverse covariance matrix. For permutation matrix estimation, we propose a relaxation technique that avoids the NP-hard combinatorial problem of order estimation. Given an ordering, a sparse Cholesky factor is estimated using a cyclic coordinatewise descent algorithm which decouples row-wise. Our framework recovers DAGs without the need for an expensive verification of the acyclicity constraint or enumeration of possible parent sets. We establish numerical convergence of the algorithm, and consistency of the Cholesky factor estimator when the order of variables is known. Through several simulated and macro-economic datasets, we study the scope and performance of the proposed methodology.

We validate the recently introduced deep learning classification adapted Delta method by a comparison with the classical Bootstrap. We show that there is a strong linear relationship between the quantified predictive epistemic uncertainty levels obtained from the two methods when applied on two LeNet-based neural network classifiers using the MNIST and CIFAR-10 datasets. Furthermore, we demonstrate that the Delta method offers a five times computation time reduction compared to the Bootstrap.

In the last decade, Deep Reinforcement Learning [1] algorithms have solved increasingly complicated problems in many different domains, spanning from video games [2] to numerous robotics applications [3], in an end-to-end fashion. Despite the success of end-to-end Reinforcement Learning, these methods suffer from low sample efficiency and usually requires lengthy and expensive training procedures to learn optimal behaviours. This problem is even more emphasized when the true state of the environment is not observable, and the observation space O or the action space A are high-dimensional. In end-to-end settings, due to the weak supervision of the reward signal, Reinforcement Learning algorithms are not enforced to learn good state representations of the environment, making the mapping observations to actions challenging to learn and interpret. State representation learning [4] methods aim at reducing the dimensionality of the observation stream by learning a mapping from the observation space O to a lower-dimensional state space S containing only the meaningful feature needed for solving a given task. By employing self-supervised auxiliary losses, it is possible to enforce optimal state representation and learn models of the underlying Markov Decision Process, or MDP. When policies are learned using the abstract or latent state-space variables, the training time is often reduced, the sample-efficiency, the robustness, and generalisation capabilities of the policies grow compared to end-to-end Reinforcement Learning [5], [6] and [7]. While the problem of state representation and observation compression has been extensively treated [4], only a few works have extended the concept of dimensionality reduction to the action space A. In this category, we find the works done in [8], [9] and [10] where low-dimensional action representations are used to improve training efficiency

Thompson sampling and other Bayesian sequential decision-making algorithms are among the most popular approaches to tackle explore/exploit trade-offs in (contextual) bandits. The choice of prior in these algorithms offers flexibility to encode domain knowledge but can also lead to poor performance when misspecified. In this paper, we demonstrate that performance degrades gracefully with misspecification. We prove that the expected reward accrued by Thompson sampling (TS) with a misspecified prior differs by at most $\tilde{\mathcal{O}}(H^2 \epsilon)$ from TS with a well specified prior, where $\epsilon$ is the total-variation distance between priors and $H$ is the learning horizon. Our bound does not require the prior to have any parametric form. For priors with bounded support, our bound is independent of the cardinality or structure of the action space, and we show that it is tight up to universal constants in the worst case. Building on our sensitivity analysis, we establish generic PAC guarantees for algorithms in the recently studied Bayesian meta-learning setting and derive corollaries for various families of priors. Our results generalize along two axes: (1) they apply to a broader family of Bayesian decision-making algorithms, including a Monte-Carlo implementation of the knowledge gradient algorithm (KG), and (2) they apply to Bayesian POMDPs, the most general Bayesian decision-making setting, encompassing contextual bandits as a special case. Through numerical simulations, we illustrate how prior misspecification and the deployment of one-step look-ahead (as in KG) can impact the convergence of meta-learning in multi-armed and contextual bandits with structured and correlated priors.

In this article we present the motivation and the core thesis towards the implementation of a Quantum Knowledge Seeking Agent (QKSA). QKSA is a general reinforcement learning agent that can be used to model classical and quantum dynamics. It merges ideas from universal artificial general intelligence, constructor theory and genetic programming to build a robust and general framework for testing the capabilities of the agent in a variety of environments. It takes the artificial life (or, animat) path to artificial general intelligence where a population of intelligent agents are instantiated to explore valid ways of modelling the perceptions. The multiplicity and survivability of the agents are defined by the fitness, with respect to the explainability and predictability, of a resource-bounded computational model of the environment. This general learning approach is then employed to model the physics of an environment based on subjective observer states of the agents. A specific case of quantum process tomography as a general modelling principle is presented. The various background ideas and a baseline formalism are discussed in this article which sets the groundwork for the implementations of the QKSA that are currently in active development.

In reinforcement learning (RL), the goal is to obtain an optimal policy, for which the optimality criterion is fundamentally important. Two major optimality criteria are average and discounted rewards, where the later is typically considered as an approximation to the former. While the discounted reward is more popular, it is problematic to apply in environments that have no natural notion of discounting. This motivates us to revisit a) the progression of optimality criteria in dynamic programming, b) justification for and complication of an artificial discount factor, and c) benefits of directly maximizing the average reward. Our contributions include a thorough examination of the relationship between average and discounted rewards, as well as a discussion of their pros and cons in RL. We emphasize that average-reward RL methods possess the ingredient and mechanism for developing the general discounting-free optimality criterion (Veinott, 1969) in RL.

In this work we theoretically and experimentally analyze Multi-Agent Advantage Actor-Critic (MA2C) and Independent Advantage Actor-Critic (IA2C), two recently proposed multi-agent reinforcement learning methods that can be applied to control traffic signals in urban areas. The two methods differ in their use of a reward calculated locally or globally and in the management of agents' communication. We analyze the methods theoretically with the framework provided by non-Markov decision processes, which provides useful insights in the analysis of the algorithms. Moreover, we analyze the efficacy and the robustness of the methods experimentally by testing them in two traffic areas in the Bologna (Italy) area, simulated by SUMO, a software tool. The experimental results indicate that MA2C achieves the best performance in the majority of cases, outperforms the alternative method considered, and displays sufficient stability during the learning process.