Given multiple non-convex objective functions and objective-specific weights, Chebyshev scalarization (CS) is a well-known approach to obtain an Exact Pareto Optimal (EPO), i.e., a solution on the Pareto front (PF) that intersects the ray defined by the inverse of the weights. First-order optimizers that use the CS formulation to find EPO solutions encounter practical problems of oscillations and stagnation that affect convergence. Moreover, when initialized with a PO solution, they do not guarantee a controlled trajectory that lies completely on the PF. These shortcomings lead to modeling limitations and computational inefficiency in multi-task learning (MTL) and multi-criteria decision-making (MCDM) methods that utilize CS for their underlying non-convex multi-objective optimization (MOO). To address these shortcomings, we design a new MOO method, EPO Search. We prove that EPO Search converges to an EPO solution and empirically illustrate its computational efficiency and robustness to initialization. When initialized on the PF, EPO Search can trace the PF and converge to the required EPO solution at a linear rate of convergence. Using EPO Search we develop new algorithms: PESA-EPO for approximating the PF in a posteriori MCDM, and GP-EPO for preference elicitation in interactive MCDM; experiments on benchmark datasets confirm their advantages over competing alternatives. EPO Search scales linearly with the number of decision variables which enables its use for training deep networks. Empirical results on real data from personalized medicine, e-commerce and hydrometeorology demonstrate the efficacy of EPO Search for deep MTL.

The goal of this thesis is to study the use of the Kantorovich-Rubinstein distance as to build a descriptor of sample complexity in classification problems. The idea is to use the fact that the Kantorovich-Rubinstein distance is a metric in the space of measures that also takes into account the geometry and topology of the underlying metric space. We associate to each class of points a measure and thus study the geometrical information that we can obtain from the Kantorovich-Rubinstein distance between those measures. We show that a large Kantorovich-Rubinstein distance between those measures allows to conclude that there exists a 1-Lipschitz classifier that classifies well the classes of points. We also discuss the limitation of the Kantorovich-Rubinstein distance as a descriptor.

Tensor data represents a multidimensional array. Regression methods based on low-rank tensor decomposition leverage structural information to reduce the parameter count. Multilinear logistic regression serves as a powerful tool for the analysis of multidimensional data. To improve its efficacy and interpretability, we present a Multilinear Sparse Logistic Regression model with $\ell_0$-constraints ($\ell_0$-MLSR). In contrast to the $\ell_1$-norm and $\ell_2$-norm, the $\ell_0$-norm constraint is better suited for feature selection. However, due to its nonconvex and nonsmooth properties, solving it is challenging and convergence guarantees are lacking. Additionally, the multilinear operation in $\ell_0$-MLSR also brings non-convexity. To tackle these challenges, we propose an Accelerated Proximal Alternating Linearized Minimization with Adaptive Momentum (APALM$^+$) method to solve the $\ell_0$-MLSR model. We provide a proof that APALM$^+$ can ensure the convergence of the objective function of $\ell_0$-MLSR. We also demonstrate that APALM$^+$ is globally convergent to a first-order critical point as well as establish convergence rate by using the Kurdyka-Lojasiewicz property. Empirical results obtained from synthetic and real-world datasets validate the superior performance of our algorithm in terms of both accuracy and speed compared to other state-of-the-art methods.

Testing the significance of a variable or group of variables $X$ for predicting a response $Y$, given additional covariates $Z$, is a ubiquitous task in statistics. A simple but common approach is to specify a linear model, and then test whether the regression coefficient for $X$ is non-zero. However, when the model is misspecified, the test may have poor power, for example when $X$ is involved in complex interactions, or lead to many false rejections. In this work we study the problem of testing the model-free null of conditional mean independence, i.e. that the conditional mean of $Y$ given $X$ and $Z$ does not depend on $X$. We propose a simple and general framework that can leverage flexible nonparametric or machine learning methods, such as additive models or random forests, to yield both robust error control and high power. The procedure involves using these methods to perform regressions, first to estimate a form of projection of $Y$ on $X$ and $Z$ using one half of the data, and then to estimate the expected conditional covariance between this projection and $Y$ on the remaining half of the data. While the approach is general, we show that a version of our procedure using spline regression achieves what we show is the minimax optimal rate in this nonparametric testing problem. Numerical experiments demonstrate the effectiveness of our approach both in terms of maintaining Type I error control, and power, compared to several existing approaches.

In this paper, we propose systematic and efficient gradient-based methods for both one-way and two-way partial AUC (pAUC) maximization that are applicable to deep learning. We propose new formulations of pAUC surrogate objectives by using the distributionally robust optimization (DRO) to define the loss for each individual positive data. We consider two formulations of DRO, one of which is based on conditional-value-at-risk (CVaR) that yields a non-smooth but exact estimator for pAUC, and another one is based on a KL divergence regularized DRO that yields an inexact but smooth (soft) estimator for pAUC. For both one-way and two-way pAUC maximization, we propose two algorithms and prove their convergence for optimizing their two formulations, respectively. Experiments demonstrate the effectiveness of the proposed algorithms for pAUC maximization for deep learning on various datasets.

Sim2Real (Simulation to Reality) techniques have gained prominence in robotic manipulation and motion planning due to their ability to enhance success rates by enabling agents to test and evaluate various policies and trajectories. In this paper, we investigate the advantages of integrating Sim2Real into robotic frameworks. We introduce the Triple Regression Sim2Real framework, which constructs a real-time digital twin. This twin serves as a replica of reality to simulate and evaluate multiple plans before their execution in real-world scenarios. Our triple regression approach addresses the reality gap by: (1) mitigating projection errors between real and simulated camera perspectives through the first two regression models, and (2) detecting discrepancies in robot control using the third regression model. Experiments on 6-DoF grasp and manipulation tasks (where the gripper can approach from any direction) highlight the effectiveness of our framework. Remarkably, with only RGB input images, our method achieves state-of-the-art success rates. This research advances efficient robot training methods and sets the stage for rapid advancements in robotics and automation.

Developing embodied agents in simulation has been a key research topic in recent years. Exciting new tasks, algorithms, and benchmarks have been developed in various simulators. However, most of them assume deaf agents in silent environments, while we humans perceive the world with multiple senses. We introduce Sonicverse, a multisensory simulation platform with integrated audio-visual simulation for training household agents that can both see and hear. Sonicverse models realistic continuous audio rendering in 3D environments in real-time. Together with a new audio-visual VR interface that allows humans to interact with agents with audio, Sonicverse enables a series of embodied AI tasks that need audio-visual perception. For semantic audio-visual navigation in particular, we also propose a new multi-task learning model that achieves state-of-the-art performance. In addition, we demonstrate Sonicverse's realism via sim-to-real transfer, which has not been achieved by other simulators: an agent trained in Sonicverse can successfully perform audio-visual navigation in real-world environments. Sonicverse is available at: https://github.com/StanfordVL/Sonicverse.

Most algorithms for representation learning and link prediction on relational data are designed for static data. However, the data to which they are applied typically evolves over time, including online social networks or interactions between users and items in recommender systems. This is also the case for graph-structured knowledge bases -- knowledge graphs -- which contain facts that are valid only for specific points in time. In such contexts, it becomes crucial to correctly identify missing links at a precise time point, i.e. the temporal prediction link task. Recently, Lacroix et al. and Sadeghian et al. proposed a solution to the problem of link prediction for knowledge graphs under temporal constraints inspired by the canonical decomposition of 4-order tensors, where they regularise the representations of time steps by enforcing temporal smoothing, i.e. by learning similar transformation for adjacent timestamps. However, the impact of the choice of temporal regularisation terms is still poorly understood. In this work, we systematically analyse several choices of temporal smoothing regularisers using linear functions and recurrent architectures. In our experiments, we show that by carefully selecting the temporal smoothing regulariser and regularisation weight, a simple method like TNTComplEx can produce significantly more accurate results than state-of-the-art methods on three widely used temporal link prediction datasets. Furthermore, we evaluate the impact of a wide range of temporal smoothing regularisers on two state-of-the-art temporal link prediction models. Our work shows that simple tensor factorisation models can produce new state-of-the-art results using newly proposed temporal regularisers, highlighting a promising avenue for future research.

A key objective in field of artificial intelligence is to develop cognitive models that can exhibit human-like intellectual capabilities. One promising approach to achieving this is through neural-symbolic systems, which combine the strengths of deep learning and symbolic reasoning. However, current approaches in this area have been limited in their combining way, generalization and interpretability. To address these limitations, we propose a general bi-level probabilistic graphical reasoning framework called GBPGR. This framework leverages statistical relational learning to effectively integrate deep learning models and symbolic reasoning in a mutually beneficial manner. In GBPGR, the results of symbolic reasoning are utilized to refine and correct the predictions made by the deep learning models. At the same time, the deep learning models assist in enhancing the efficiency of the symbolic reasoning process. Through extensive experiments, we demonstrate that our approach achieves high performance and exhibits effective generalization in both transductive and inductive tasks.

The emergence of human-like abilities of AI systems for content generation in domains such as text, audio, and vision has prompted the development of classifiers to determine whether content originated from a human or a machine. Implicit in these efforts is an assumption that the generation properties of a human are different from that of the machine. In this work, we provide a framework in the language of statistical pattern recognition that quantifies the difference between the distributions of human and machine-generated content conditioned on an evaluation context. We describe current methods in the context of the framework and demonstrate how to use the framework to evaluate the progression of generative models towards human-like capabilities, among many axes of analysis.