The back-propagation algorithm has long been the de-facto standard in optimizing weights and biases in neural networks, particularly in cutting-edge deep learning models. Its widespread adoption in fields like natural language processing, computer vision, and remote sensing has revolutionized automation in various tasks. The popularity of back-propagation stems from its ability to achieve outstanding performance in tasks such as classification, detection, and segmentation. Nevertheless, back-propagation is not without its limitations, encompassing sensitivity to initial conditions, vanishing gradients, overfitting, and computational complexity. The recent introduction of a forward-forward algorithm (FFA), which computes local goodness functions to optimize network parameters, alleviates the dependence on substantial computational resources and the constant need for architectural scaling. This study investigates the application of FFA for hyperspectral image classification. Experimental results and comparative analysis are provided with the use of the traditional back-propagation algorithm. Preliminary results show the potential behind FFA and its promises.

Federated learning (FL) is an emerging paradigm to train model with distributed data from numerous Internet of Things (IoT) devices. It inherently assumes a uniform capacity among participants. However, due to different conditions such as differing energy budgets or executing parallel unrelated tasks, participants have diverse computational resources in practice. Participants with insufficient computation budgets must plan for the use of restricted computational resources appropriately, otherwise they would be unable to complete the entire training procedure, resulting in model performance decline. To address this issue, we propose a strategy for estimating local models without computationally intensive iterations. Based on it, we propose Computationally Customized Federated Averaging (CC-FedAvg), which allows participants to determine whether to perform traditional local training or model estimation in each round based on their current computational budgets. Both theoretical analysis and exhaustive experiments indicate that CC-FedAvg has the same convergence rate and comparable performance as FedAvg without resource constraints. Furthermore, CC-FedAvg can be viewed as a computation-efficient version of FedAvg that retains model performance while considerably lowering computation overhead.

We study sample efficient reinforcement learning (RL) under the general framework of interactive decision making, which includes Markov decision process (MDP), partially observable Markov decision process (POMDP), and predictive state representation (PSR) as special cases. Toward finding the minimum assumption that empowers sample efficient learning, we propose a novel complexity measure, generalized eluder coefficient (GEC), which characterizes the fundamental tradeoff between exploration and exploitation in online interactive decision making. In specific, GEC captures the hardness of exploration by comparing the error of predicting the performance of the updated policy with the in-sample training error evaluated on the historical data. We show that RL problems with low GEC form a remarkably rich class, which subsumes low Bellman eluder dimension problems, bilinear class, low witness rank problems, PO-bilinear class, and generalized regular PSR, where generalized regular PSR, a new tractable PSR class identified by us, includes nearly all known tractable POMDPs and PSRs. Furthermore, in terms of algorithm design, we propose a generic posterior sampling algorithm, which can be implemented in both model-free and model-based fashion, under both fully observable and partially observable settings. The proposed algorithm modifies the standard posterior sampling algorithm in two aspects: (i) we use an optimistic prior distribution that biases towards hypotheses with higher values and (ii) a loglikelihood function is set to be the empirical loss evaluated on the historical data, where the choice of loss function supports both model-free and model-based learning. We prove that the proposed algorithm is sample efficient by establishing a sublinear regret upper bound in terms of GEC. In summary, we provide a new and unified understanding of both fully observable and partially observable RL.

This paper addresses the problem of traversing through unknown, tilted, and narrow gaps for quadrotors using Deep Reinforcement Learning (DRL). Previous learning-based methods relied on accurate knowledge of the environment, including the gap's pose and size. In contrast, we integrate onboard sensing and detect the gap from a single onboard camera. The training problem is challenging for two reasons: a precise and robust whole-body planning and control policy is required for variable-tilted and narrow gaps, and an effective Sim2Real method is needed to successfully conduct real-world experiments. To this end, we propose a learning framework for agile gap traversal flight, which successfully trains the vehicle to traverse through the center of the gap at an approximate attitude to the gap with aggressive tilted angles. The policy trained only in a simulation environment can be transferred into different domains with fine-tuning while maintaining the success rate. Our proposed framework, which integrates onboard sensing and a neural network controller, achieves a success rate of 84.51% in real-world experiments, with gap orientations up to 60deg. To the best of our knowledge, this is the first paper that performs the learning-based variable-tilted narrow gap traversal flight in the real world, without prior knowledge of the environment.

The global push to advance Carbon Capture and Sequestration initiatives and green energy solutions, such as geothermal, have thrust new demands upon the current state-of-the-art subsurface fluid simulators. The requirement to be able to simulate a large order of reservoir states simultaneously, in a short period of time, has opened the door of opportunity for the application of machine learning techniques for surrogate modelling. We propose a novel physics-informed and boundary condition-aware Localized Learning method which extends the Embed-to-Control (E2C) and Embed-to-Control and Observe (E2CO) models to learn local representations of global state variables in an Advection-Diffusion Reaction system. Trained on reservoir simulation data, we show that our model is able to predict future states of the system, for a given set of controls, to a great deal of accuracy with only a fraction of the available information. It hence reduces training times significantly compared to the original E2C and E2CO models, lending to its benefit in application to optimal control problems.

Climate change is the defining issue of our time, and we are at a defining moment. Various interest groups, social movement organizations, and individuals engage in collective action on this issue on social media. In addition, issue advocacy campaigns on social media often arise in response to ongoing societal concerns, especially those faced by energy industries. Our goal in this paper is to analyze how those industries, their advocacy group, and climate advocacy group use social media to influence the narrative on climate change. In this work, we propose a minimally supervised model soup [57] approach combined with messaging themes to identify the stances of climate ads on Facebook. Finally, we release our stance dataset, model, and set of themes related to climate campaigns for future work on opinion mining and the automatic detection of climate change stances.

Bayesian Optimization (BO) is a class of surrogate-based, sample-efficient algorithms for optimizing black-box problems with small evaluation budgets. The BO pipeline itself is highly configurable with many different design choices regarding the initial design, surrogate model, and acquisition function (AF). Unfortunately, our understanding of how to select suitable components for a problem at hand is very limited. In this work, we focus on the definition of the AF, whose main purpose is to balance the trade-off between exploring regions with high uncertainty and those with high promise for good solutions. We propose Self-Adjusting Weighted Expected Improvement (SAWEI), where we let the exploration-exploitation trade-off self-adjust in a data-driven manner, based on a convergence criterion for BO. On the noise-free black-box BBOB functions of the COCO benchmarking platform, our method exhibits a favorable any-time performance compared to handcrafted baselines and serves as a robust default choice for any problem structure. The suitability of our method also transfers to HPOBench. With SAWEI, we are a step closer to on-the-fly, data-driven, and robust BO designs that automatically adjust their sampling behavior to the problem at hand.

Goal-conditioned hierarchical reinforcement learning (GCHRL) decomposes long-horizon tasks into sub-tasks through a hierarchical framework and it has demonstrated promising results across a variety of domains. However, the high-level policy's action space is often excessively large, presenting a significant challenge to effective exploration and resulting in potentially inefficient training. Moreover, the dynamic variability of the low-level policy introduces non-stationarity to the high-level state transition function, significantly impeding the learning of the high-level policy. In this paper, we design a measure of prospect for subgoals by planning in the goal space based on the goal-conditioned value function. Building upon the measure of prospect, we propose a landmark-guided exploration strategy by integrating the measures of prospect and novelty which aims to guide the agent to explore efficiently and improve sample efficiency. To address the non-stationarity arising from the dynamic changes of the low-level policy, we apply a state-specific regularization to the learning of low-level policy, which facilitates stable learning of the hierarchical policy. The experimental results demonstrate that our proposed exploration strategy significantly outperforms the baseline methods across multiple tasks.

We propose a novel probabilistically robust controller for the guidance of an unmanned aerial vehicle (UAV) in coverage planning missions, which can simultaneously optimize both the UAV's motion, and camera control inputs for the 3D coverage of a given object of interest. Specifically, the coverage planning problem is formulated in this work as an optimal control problem with logical constraints to enable the UAV agent to jointly: a) select a series of discrete camera field-of-view states which satisfy a set of coverage constraints, and b) optimize its motion control inputs according to a specified mission objective. We show how this hybrid optimal control problem can be solved with standard optimization tools by converting the logical expressions in the constraints into equality/inequality constraints involving only continuous variables. Finally, probabilistic robustness is achieved by integrating the unscented transformation to the proposed controller, thus enabling the design of robust open-loop coverage plans which take into account the future posterior distribution of the UAV's state inside the planning horizon.

Robot programming tools ranging from inverse kinematics (IK) to model predictive control (MPC) are most often described as constrained optimization problems. Even though there are currently many commercially-available second-order solvers, robotics literature recently focused on efficient implementations and improvements over these solvers for real-time robotic applications. However, most often, these implementations stay problem-specific and are not easy to access or implement, or do not exploit the geometric aspect of the robotics problems. In this work, we propose to solve these problems using a fast, easy-to-implement first-order method that fully exploits the geometric constraints via Euclidean projections, called Augmented Lagrangian Spectral Projected Gradient Descent (ALSPG). We show that 1. using projections instead of full constraints and gradients improves the performance of the solver and 2. ALSPG stays competitive to the standard second-order methods such as iLQR in the unconstrained case. We showcase these results with IK and motion planning problems on simulated examples and with an MPC problem on a 7-axis manipulator experiment.