We present a novel learning framework to solve the transport-and-packing (TAP) problem in 3D. It constitutes a full solution pipeline from partial observations of input objects via RGBD sensing and recognition to final box placement, via robotic motion planning, to arrive at a compact packing in a target container. The technical core of our method is a neural network for TAP, trained via reinforcement learning (RL), to solve the NP-hard combinatorial optimization problem. Our network simultaneously selects an object to pack and determines the final packing location, based on a judicious encoding of the continuously evolving states of partially observed source objects and available spaces in the target container, using separate encoders both enabled with attention mechanisms. The encoded feature vectors are employed to compute the matching scores and feasibility masks of different pairings of box selection and available space configuration for packing strategy optimization. Extensive experiments, including ablation studies and physical packing execution by a real robot (Universal Robot UR5e), are conducted to evaluate our method in terms of its design choices, scalability, generalizability, and comparisons to baselines, including the most recent RL-based TAP solution. We also contribute the first benchmark for TAP which covers a variety of input settings and difficulty levels.

The ability to learn good representations of states is essential for solving large reinforcement learning problems, where exploration, generalization, and transfer are particularly challenging. The Laplacian representation is a promising approach to address these problems by inducing intrinsic rewards for temporally-extended action discovery and reward shaping, and informative state encoding. To obtain the Laplacian representation one needs to compute the eigensystem of the graph Laplacian, which is often approximated through optimization objectives compatible with deep learning approaches. These approximations, however, depend on hyperparameters that are impossible to tune efficiently, converge to arbitrary rotations of the desired eigenvectors, and are unable to accurately recover the corresponding eigenvalues. In this paper we introduce a theoretically sound objective and corresponding optimization algorithm for approximating the Laplacian representation. Our approach naturally recovers both the true eigenvectors and eigenvalues while eliminating the hyperparameter dependence of previous approximations. We provide theoretical guarantees for our method and we show that those results translate empirically into robust learning across multiple environments.

Our ultimate goal is to build robust policies for robots that assist people. What makes this hard is that people can behave unexpectedly at test time, potentially interacting with the robot outside its training distribution and leading to failures. Even just measuring robustness is a challenge. Adversarial perturbations are the default, but they can paint the wrong picture: they can correspond to human motions that are unlikely to occur during natural interactions with people. A robot policy might fail under small adversarial perturbations but work under large natural perturbations. We propose that capturing robustness in these interactive settings requires constructing and analyzing the entire natural-adversarial frontier: the Pareto-frontier of human policies that are the best trade-offs between naturalness and low robot performance. We introduce RIGID, a method for constructing this frontier by training adversarial human policies that trade off between minimizing robot reward and acting human-like (as measured by a discriminator). On an Assistive Gym task, we use RIGID to analyze the performance of standard collaborative Reinforcement Learning, as well as the performance of existing methods meant to increase robustness. We also compare the frontier RIGID identifies with the failures identified in expert adversarial interaction, and with naturally-occurring failures during user interaction. Overall, we find evidence that RIGID can provide a meaningful measure of robustness predictive of deployment performance, and uncover failure cases in human-robot interaction that are difficult to find manually. https://ood-human.github.io.

The discovery of therapeutic molecules is fundamentally a multi-objective optimization problem. One formulation of the problem is to identify molecules that simultaneously exhibit strong binding affinity for a target protein, minimal off-target interactions, and suitable pharmacokinetic properties. Inspired by prior work that uses active learning to accelerate the identification of strong binders, we implement multi-objective Bayesian optimization to reduce the computational cost of multi-property virtual screening and apply it to the identification of ligands predicted to be selective based on docking scores to on- and off-targets. We demonstrate the superiority of Pareto optimization over scalarization across three case studies. Further, we use the developed optimization tool to search a virtual library of over 4M molecules for those predicted to be selective dual inhibitors of EGFR and IGF1R, acquiring 100% of the molecules that form the library's Pareto front after exploring only 8% of the library. This workflow and associated open source software can reduce the screening burden of molecular design projects and is complementary to research aiming to improve the accuracy of binding predictions and other molecular properties.

Contact-rich manipulation is common in a wide range of robotic applications, including assembly [1, 2, 3, 4, 5, 6, 7, 8], object pivoting [9, 10, 11], grasping[12, 13, 14], and pushing [15, 16]. To accomplish these tasks, robots need to learn both the manipulation trajectory and the force control parameters. The manipulation trajectory guides the robot toward completing the task while physically engaging with the environment, whereas the force control parameters regulate the contact force. Incorrect control parameters can lead to oscillations and excessive contact forces that may damage the robot or the environment. Past works have tackled the contact-rich skill-learning problem in different ways. First, the majority of previous works [2, 10, 6, 7, 3, 17, 9, 18, 19] focus on learning the manipulation trajectories and rely on human experts to manually tune force control parameters. While this simplification has demonstrated remarkable performance in many applications, letting human labor tune control parameters is still inconvenient. Furthermore, the tuned parameter for one task may not generalize well to other task settings with different kinematic or dynamic properties. For example, assembly tasks with different clearances will require different control parameters.

To date, power electronics parameter design tasks are usually tackled using detailed optimization approaches with detailed simulations or using brute force grid search grid search with very fast simulations. A new method, named "Continuously Adapting Random Sampling" (CARS) is proposed, which provides a continuous method in between. This allows for very fast, and / or large amounts of simulations, but increasingly focuses on the most promising parameter ranges. Inspirations are drawn from multi-armed bandit research and lead to prioritized sampling of sub-domains in one high-dimensional parameter tensor. Performance has been evaluated on three exemplary power electronic use-cases, where resulting designs appear competitive to genetic algorithms, but additionally allow for highly parallelizable simulation, as well as continuous progression between explorative and exploitative settings.

This paper considers federated learning (FL) with constraints, where the central server and all local clients collectively minimize a sum of convex local objective functions subject to global and local convex conic constraints. To train the model without moving local data from clients to the central server, we propose an FL framework in which each local client performs multiple updates using the local objective and local constraint, while the central server handles the global constraint and performs aggregation based on the updated local models. In particular, we develop a proximal augmented Lagrangian (AL) based algorithm for FL with global and local convex conic constraints. The subproblems arising in this algorithm are solved by an inexact alternating direction method of multipliers (ADMM) in a federated fashion. Under a local Lipschitz condition and mild assumptions, we establish the worst-case complexity bounds of the proposed algorithm for finding an approximate KKT solution. To the best of our knowledge, this work proposes the first algorithm for FL with global and local constraints. Our numerical experiments demonstrate the practical advantages of our algorithm in performing Neyman-Pearson classification and enhancing model fairness in the context of FL.

Federated learning (FL), as an emerging distributed machine learning paradigm, allows a mass of edge devices to collaboratively train a global model while preserving privacy. In this tutorial, we focus on FL via over-the-air computation (AirComp), which is proposed to reduce the communication overhead for FL over wireless networks at the cost of compromising in the learning performance due to model aggregation error arising from channel fading and noise. We first provide a comprehensive study on the convergence of AirComp-based FedAvg (AirFedAvg) algorithms under both strongly convex and non-convex settings with constant and diminishing learning rates in the presence of data heterogeneity. Through convergence and asymptotic analysis, we characterize the impact of aggregation error on the convergence bound and provide insights for system design with convergence guarantees. Then we derive convergence rates for AirFedAvg algorithms for strongly convex and non-convex objectives. For different types of local updates that can be transmitted by edge devices (i.e., local model, gradient, and model difference), we reveal that transmitting local model in AirFedAvg may cause divergence in the training procedure. In addition, we consider more practical signal processing schemes to improve the communication efficiency and further extend the convergence analysis to different forms of model aggregation error caused by these signal processing schemes. Extensive simulation results under different settings of objective functions, transmitted local information, and communication schemes verify the theoretical conclusions.

In computational chemistry, crystal structure prediction (CSP) is an optimization problem that involves discovering the lowest energy stable crystal structure for a given chemical formula. This problem is challenging as it requires discovering globally optimal designs with the lowest energies on complex manifolds. One approach to tackle this problem involves building simulators based on density functional theory (DFT) followed by running search in simulation, but these simulators are painfully slow. In this paper, we study present and study an alternate, data-driven approach to crystal structure prediction: instead of directly searching for the most stable structures in simulation, we train a surrogate model of the crystal formation energy from a database of existing crystal structures, and then optimize this model with respect to the parameters of the crystal structure. This surrogate model is trained to be conservative so as to prevent exploitation of its errors by the optimizer. To handle optimization in the non-Euclidean space of crystal structures, we first utilize a state-of-the-art graph diffusion auto-encoder (CD-VAE) to convert a crystal structure into a vector-based search space and then optimize a conservative surrogate model of the crystal energy, trained on top of this vector representation. We show that our approach, dubbed LCOMs (latent conservative objective models), performs comparably to the best current approaches in terms of success rate of structure prediction, while also drastically reducing computational cost.

Nature-inspired metaheuristic algorithms are important components of artificial intelligence, and are increasingly used across disciplines to tackle various types of challenging optimization problems. We apply a newly proposed nature-inspired metaheuristic algorithm called competitive swarm optimizer with mutated agents (CSO-MA) and demonstrate its flexibility and out-performance relative to its competitors in a variety of optimization problems in the statistical sciences. In particular, we show the algorithm is efficient and can incorporate various cost structures or multiple user-specified nonlinear constraints. Our applications include (i) finding maximum likelihood estimates of parameters in a single cell generalized trend model to study pseudotime in bioinformatics, (ii) estimating parameters in a commonly used Rasch model in education research, (iii) finding M-estimates for a Cox regression in a Markov renewal model and (iv) matrix completion to impute missing values in a two compartment model. In addition we discuss applications to (v) select variables optimally in an ecology problem and (vi) design a car refueling experiment for the auto industry using a logistic model with multiple interacting factors.