Everyday activities like using a GPS to map out the best driving route or translating a document consume energy, water and mineral resources -- lots of it. These applications run in the cloud, a nebulous term for the millions of powerful computers in vast data centers worldwide. Mobile applications depend on legions of computers to store trillions of data and perform split-second operations (e.g. Estimates of the energy consumption of data centers range between 1-2% of total global consumption. All signs indicate that data center energy consumption is about to skyrocket.

Inspired by the numerical solution of ordinary differential equations, in this paper we propose a novel Reservoir Computing (RC) model, called the Euler State Network (EuSN). The presented approach makes use of forward Euler discretization and antisymmetric recurrent matrices to design reservoir dynamics that are both stable and non-dissipative by construction. Our mathematical analysis shows that the resulting model is biased towards a unitary effective spectral radius and zero local Lyapunov exponents, intrinsically operating near to the edge of stability. Experiments on long-term memory tasks show the clear superiority of the proposed approach over standard RC models in problems requiring effective propagation of input information over multiple time-steps. Furthermore, results on time-series classification benchmarks indicate that EuSN is able to match (or even exceed) the accuracy of trainable Recurrent Neural Networks, while retaining the training efficiency of the RC family, resulting in up to $\approx$ 490-fold savings in computation time and $\approx$ 1750-fold savings in energy consumption.

Deep learning has had remarkable success in robotic perception, but its data-centric nature suffers when it comes to generalizing to ever-changing environments. By contrast, physics-based optimization generalizes better, but it does not perform as well in complicated tasks due to the lack of high-level semantic information and reliance on manual parametric tuning. To take advantage of these two complementary worlds, we present PyPose: a robotics-oriented, PyTorch-based library that combines deep perceptual models with physics-based optimization. PyPose's architecture is tidy and well-organized, it has an imperative style interface and is efficient and user-friendly, making it easy to integrate into real-world robotic applications. Besides, it supports parallel computing of any order gradients of Lie groups and Lie algebras and $2^{\text{nd}}$-order optimizers, such as trust region methods. Experiments show that PyPose achieves more than $10\times$ speedup in computation compared to the state-of-the-art libraries. To boost future research, we provide concrete examples for several fields of robot learning, including SLAM, planning, control, and inertial navigation.

Despite its importance for federated learning, continuous learning and many other applications, on-device training remains an open problem for EdgeAI. The problem stems from the large number of operations (e.g., floating point multiplications and additions) and memory consumption required during training by the back-propagation algorithm. Consequently, in this paper, we propose a new gradient filtering approach which enables on-device CNN model training. More precisely, our approach creates a special structure with fewer unique elements in the gradient map, thus significantly reducing the computational complexity and memory consumption of back propagation during training. Extensive experiments on image classification and semantic segmentation with multiple CNN models (e.g., MobileNet, DeepLabV3, UPerNet) and devices (e.g., Raspberry Pi and Jetson Nano) demonstrate the effectiveness and wide applicability of our approach. For example, compared to SOTA, we achieve up to 19$\times$ speedup and 77.1% memory savings on ImageNet classification with only 0.1% accuracy loss. Finally, our method is easy to implement and deploy; over 20$\times$ speedup and 90% energy savings have been observed compared to highly optimized baselines in MKLDNN and CUDNN on NVIDIA Jetson Nano. Consequently, our approach opens up a new direction of research with a huge potential for on-device training.

In-Context Learning (ICL), which formulates target tasks as prompt completion conditioned on in-context demonstrations, has become the prevailing utilization of LLMs. In this paper, we first disclose an actual predicament for this typical usage that it can not scale up with training data due to context length restriction. Besides, existing works have shown that ICL also suffers from various biases and requires delicate calibration treatment. To address both challenges, we advocate a simple and effective solution, $k$NN Prompting, which first queries LLM with training data for distributed representations, then predicts test instances by simply referring to nearest neighbors. We conduct comprehensive experiments to demonstrate its two-fold superiority: 1) Calibration-Free: $k$NN Prompting does not directly align LLM output distribution with task-specific label space, instead leverages such distribution to align test and training instances. It significantly outperforms state-of-the-art calibration-based methods under comparable few-shot scenario. 2) Beyond-Context: $k$NN Prompting can further scale up effectively with as many training data as are available, continually bringing substantial improvements. The scaling trend holds across 10 orders of magnitude ranging from 2 shots to 1024 shots as well as different LLMs scales ranging from 0.8B to 30B. It successfully bridges data scaling into model scaling, and brings new potentials for the gradient-free paradigm of LLM deployment. Code is publicly available.

We present a near real-time method for 6-DoF tracking of an unknown object from a monocular RGBD video sequence, while simultaneously performing neural 3D reconstruction of the object. Our method works for arbitrary rigid objects, even when visual texture is largely absent. The object is assumed to be segmented in the first frame only. No additional information is required, and no assumption is made about the interaction agent. Key to our method is a Neural Object Field that is learned concurrently with a pose graph optimization process in order to robustly accumulate information into a consistent 3D representation capturing both geometry and appearance. A dynamic pool of posed memory frames is automatically maintained to facilitate communication between these threads. Our approach handles challenging sequences with large pose changes, partial and full occlusion, untextured surfaces, and specular highlights. We show results on HO3D, YCBInEOAT, and BEHAVE datasets, demonstrating that our method significantly outperforms existing approaches. Project page: https://bundlesdf.github.io

The increasing importance of carbon capture technologies for deployment in remediating CO2 emissions, and thus the necessity to improve capture materials to allow scalability and efficiency, faces the challenge of materials development, which can require substantial costs and time. Machine learning offers a promising method for reducing the time and resource burdens of materials development through efficient correlation of structure-property relationships to allow down-selection and focusing on promising candidates. Towards demonstrating this, we have developed an end-to-end "discovery cycle" to select new aqueous amines compatible with the commercially viable acid gas scrubbing carbon capture. We combine a simple, rapid laboratory assay for CO2 absorption with a machine learning based molecular fingerprinting model approach. The prediction process shows 60% accuracy against experiment for both material parameters and 80% for a single parameter on an external test set. The discovery cycle determined several promising amines that were verified experimentally, and which had not been applied to carbon capture previously. In the process we have compiled a large, single-source data set for carbon capture amines and produced an open source machine learning tool for the identification of amine molecule candidates (https://github.com/IBM/Carbon-capture-fingerprint-generation).

In many areas of the observational and experimental sciences data is scarce. Data observation in high-energy astrophysics is disrupted by celestial occlusions and limited telescope time while data derived from laboratory experiments in synthetic chemistry and materials science is time and cost-intensive to collect. On the other hand, knowledge about the data-generation mechanism is often available in the sciences, such as the measurement error of a piece of laboratory apparatus. Both characteristics, small data and knowledge of the underlying physics, make Gaussian processes (GPs) ideal candidates for fitting such datasets. GPs can make predictions with consideration of uncertainty, for example in the virtual screening of molecules and materials, and can also make inferences about incomplete data such as the latent emission signature from a black hole accretion disc. Furthermore, GPs are currently the workhorse model for Bayesian optimisation, a methodology foreseen to be a guide for laboratory experiments in scientific discovery campaigns. The first contribution of this thesis is to use GP modelling to reason about the latent emission signature from the Seyfert galaxy Markarian 335, and by extension, to reason about the applicability of various theoretical models of black hole accretion discs. The second contribution is to extend the GP framework to molecular and chemical reaction representations and to provide an open-source software library to enable the framework to be used by scientists. The third contribution is to leverage GPs to discover novel and performant photoswitch molecules. The fourth contribution is to introduce a Bayesian optimisation scheme capable of modelling aleatoric uncertainty to facilitate the identification of material compositions that possess intrinsic robustness to large scale fabrication processes.

This paper develops a Deep Reinforcement Learning (DRL)-agent for navigation and control of autonomous surface vessels (ASV) on inland waterways. Spatial restrictions due to waterway geometry and the resulting challenges, such as high flow velocities or shallow banks, require controlled and precise movement of the ASV. A state-of-the-art bootstrapped Q-learning algorithm in combination with a versatile training environment generator leads to a robust and accurate rudder controller. To validate our results, we compare the path-following capabilities of the proposed approach to a vessel-specific PID controller on real-world river data from the lower- and middle Rhine, indicating that the DRL algorithm could effectively prove generalizability even in never-seen scenarios while simultaneously attaining high navigational accuracy.

Energetic materials (EM) such as propellants, explosives, and pyrotechnics are key components in many military and civilian applications. EMs are composites of organic crystals, plasticizers, metals, and other inclusions, forming complex microstructural morphologies, which strongly influence the properties and performance characteristics of these materials (1). For instance, the sensitivity to impact and shock loading--one of the key performance parameters for the design of safe and reliable EMs--is strongly influenced by their microstructures (2-4). Voids, cracks, and interfaces in EM microstructures are potential sites for energy localization, i.e., the formation of hightemperature regions called "hotspots" (5-8). Such hotspots are considered to be critical if they grow and produce steady deflagration fronts (9). If a sufficient number of such critical hotspots are generated in the microstructure, chemical energy release can be rapid enough to couple with the incident shock wave, initiating a detonation. Therefore, microstructural features localize energy release at hotspots and shock-microstructure interactions can lead to a shock-to-detonation transition in EMs. 1