Motion models for robust 3D human body tracking

In this work, we propose new ways to learn pose and motion priors models and show that they can be used to increase the performance of 3D body tracking algorithms, resulting in very realistic motions under very challenging conditions. We first explored an approach to 3D people tracking that combines learned motion models and deterministic optimization. The tracking problem is formulated as the minimization of a differentiable criterion whose differential structure is rich enough for optimization to be accomplished via hill-climbing. This avoids the computational expense of Monte Carlo methods, while yielding very good results under challenging conditions. To demonstrate the generality of the approach we show that we can learn and track cyclic motions such as walking and running, as well as acyclic motions such as a golf swing. We also show results from both monocular and multi-camera tracking. Finally, we provide results with a motion model learned from multiple activities, and show how these models can be used for recognition and motion generation. The major limitation of these linear motion models is that they required many noiseless, segmented and time warped examples to create a complete database with good generalization properties. We therefore investigated more complex non-linear statistical techniques. We advocate the use of Scaled Gaussian Process Latent Variable Models (SGPLVM) to learn prior models of 3D human pose. The SGPLVM simultaneously optimizes a low-dimensional embedding of the high-dimensional pose data and a density function that both gives higher probability to points close to training data and provides a nonlinear probabilistic mapping from the low-dimensional latent space to the full-dimensional pose space. The SGPLVM is a natural choice when only small amounts of training data are available. We demonstrate our approach with two distinct motions, golfing and walking. We show that the SGPLVM sufficiently constrains the problem such that tracking can be accomplished with straightforward deterministic optimization. However, in the presence of very noisy or missing data, for example due to occlusions, the simplistic second order Markov model we use is not realistic enough to sufficiently constrain the algorithm. Moreover, when learning models that contain stylistic diversity, from different people or from the same person performing an activity multiple times, the SGPLVM results in models whose latent trajectories are not smooth, and are therefore not suited for hill climbing tracking. Finally, we present a more powerfull approach based on the Gaussian Process Dynamical Models (GPDMs) that combines the strengths of the two previous ones. We advocate the use of GPDMs for learning human pose and motion priors. A GPDM provides a low-dimensional embedding of human motion data, with a density function that gives higher probability to poses and motions close to the training data. With Bayesian model averaging a GPDM can be learned from relatively small amounts of data, and it generalizes gracefully to motions outside the training set. Here we modify the GPDM to permit learning from motions with significant stylistic variation. The resulting priors are effective for tracking a range of human walking styles, despite weak and noisy image measurements and significant occlusions.

Robust Adaptive Decision Making: Bayesian Optimization and Beyond

Ilija Bogunovic

The central task in many interactive machine learning systems can be formalized as the sequential optimization of a black-box function. Bayesian optimization (BO) is a powerful model-based framework for \emph{adaptive} experimentation, where the primary goal is the optimization of the black-box function via sequentially chosen decisions. In many real-world tasks, it is essential for the decisions to be \emph{robust} against, e.g., adversarial failures and perturbations, dynamic and time-varying phenomena, a mismatch between simulations and reality, etc. Under such requirements, the standard methods and BO algorithms become inadequate. In this dissertation, we consider four research directions with the goal of enhancing robust and adaptive decision making in BO and associated problems. First, we study the related problem of level-set estimation (LSE) with Gaussian Processes (GPs). While in BO the goal is to find a maximizer of the unknown function, in LSE one seeks to find all "sufficiently good" solutions. We propose an efficient confidence-bound based algorithm that treats BO and LSE in a unified fashion. It is effective in settings that are non-trivial to incorporate into existing algorithms, including cases with pointwise costs, heteroscedastic noise, and multi-fidelity setting. Our main result is a general regret guarantee that covers these aspects. Next, we consider GP optimization with robustness requirement: An adversary may perturb the returned design, and so we seek to find a robust maximizer in the case this occurs. This requirement is motivated by, e.g., settings where the functions during optimization and implementation stages are different. We propose a novel robust confidence-bound based algorithm. The rigorous regret guarantees for this algorithm are established and complemented with an algorithm-independent lower bound. We experimentally demonstrate that our robust approach consistently succeeds in finding a robust maximizer while standard BO methods fail. We then investigate the problem of GP optimization in which the reward function varies with time. The setting is motivated by many practical applications in which the function to be optimized is not static. We model the unknown reward function via a GP whose evolution obeys a simple Markov model. Two confidence-bound based algorithms with the ability to "forget" about old data are proposed. We obtain regret bounds for these algorithms that jointly depend on the time horizon and the rate at which the function varies. Finally, we consider the maximization of a set function subject to a cardinality constraint

k

in the case a number of items

\tau

from the returned set may be removed. One notable application is in batch BO where we need to select experiments to run, but some of them can fail. Our focus is on the worst-case adversarial setting, and we consider both \emph{submodular} (i.e., satisfies a natural notion of diminishing returns) and \emph{non-submodular} objectives. We propose robust algorithms that achieve constant-factor approximation guarantees. In the submodular case, the result on the maximum number of allowed removals is improved to

\tau = o(k)

in comparison to the previously known

\tau=o(\sqrt{k})

. In the non-submodular case, we obtain new guarantees in the support selection and batch BO tasks. We empirically demonstrate the robust performance of our algorithms in these, as well as, in data summarization and influence maximization tasks.

EPFL2019

Optimization methods for collaborative learning

Sai Praneeth Reddy Karimireddy

A traditional machine learning pipeline involves collecting massive amounts of data centrally on a server and training models to fit the data. However, increasing concerns about the privacy and security of user's data, combined with the sheer growth in the data sizes has incentivized looking beyond such traditional centralized approaches. Collaborative learning (which encompasses distributed, federated, and decentralized learning) proposes instead for a network of data holders to collaborate together to train models without transmitting any data. This new paradigm minimizes data exposure, but inherently faces some fundamental challenges. In this thesis, we bring to bear the framework of stochastic optimization to formalize and develop new algorithms for these challenges. This serves not only to develop novel solutions, but also to test the utility of the optimization lens in modern deep learning.We study three fundamental problems. Firstly, collaborative training replaces a one-time transmission of raw data with repeated rounds of communicating partially trained models. However, this quickly runs against bandwidth constraints when dealing with large models. We propose to solve this bandwidth constraint using compressed communication. Next, collaborative training leverages the computation power of the data holders directly. However, this is not as reliable as using a data center with only a subset of them available at any given time. Thus, we require new algorithms which can efficiently utilize unreliable local computation of the data holders. Finally, collaborative training allows any data holder to participate in the training process, without being able to inspect their data or local computation. This may potentially open the system to malicious or faulty agents who seek to derail the training. We develop algorithms with Byzantine robustness which are guaranteed to be resilient to such attackers.

EPFL2021

Ensembles for sequence learning

Christos Dimitrakakis

This thesis explores the application of ensemble methods to sequential learning tasks. The focus is on the development and the critical examination of new methods or novel applications of existing methods, with emphasis on supervised and reinforcement learning problems. In both types of problems, even after having observed a certain amount of data, we are often faced with uncertainty as to which hypothesis is correct among all the possible ones. However, in many methods for both supervised and for reinforcement learning problems this uncertainty is ignored, in the sense that there is a single solution selected out of the whole of the hypothesis space. Apart from the classical solution of analytical Bayesian formulations, ensemble methods offer an alternative approach to representing this uncertainty. This is done simply through maintaining a set of alternative hypotheses. The sequential supervised problem considered is that of automatic speech recognition using hidden Markov models. The application of ensemble methods to the problem represents a challenge in itself, since most such methods can not be readily adapted to sequential learning tasks. This thesis proposes a number of different approaches for applying ensemble methods to speech recognition and develops methods for effective training of phonetic mixtures with or without access to phonetic alignment data. Furthermore, the notion of expected loss is introduced for integrating probabilistic models with the boosting approach. In some cases substantial improvements over the baseline system are obtained. In reinforcement learning problems the goal is to act in such a way as to maximise future reward in a given environment. In such problems uncertainty becomes important since neither the environment nor the distribution of rewards that result from each action are known. This thesis presents novel algorithms for acting nearly optimally under uncertainty based on theoretical considerations. Some ensemble-based representations of uncertainty (including a fully Bayesian model) are developed and tested on a few simple tasks resulting in performance comparable with the state of the art. The thesis also draws some parallels between a proposed representation of uncertainty based on gradient-estimates and on"prioritised sweeping" and between the application of reinforcement learning to controlling an ensemble of classifiers and classical supervised ensemble learning methods.

EPFL2007