Search

Deep state-space models (Deep SSMs) have shown capabilities for in-context learning on autoregressive tasks, similar to transformers. However, the architectural requirements and mechanisms enabling this in recurrent networks remain unclear. This study demonstrates that state-space model architectures can perform gradient-based learning and use it for in-context learning. We prove that a single structured state-space model layer, augmented with local self-attention, can reproduce the outputs of an implicit linear model with least squares loss after one step of gradient descent. Our key insight is that the diagonal linear recurrent layer can act as a gradient accumulator, which can be `applied' to the parameters of the implicit regression model. We validate our construction by training randomly initialized augmented SSMs on simple linear regression tasks. The empirically optimized parameters match the theoretical ones, obtained analytically from the implicit model construction. Extensions to multi-step linear and non-linear regression yield consistent results. The constructed SSM encompasses features of modern deep state-space models, with the potential for scalable training and effectiveness even in general tasks. The theoretical construction elucidates the role of local self-attention and multiplicative interactions in recurrent architectures as the key ingredients for enabling the expressive power typical of foundation models., Comment: 16 pages, 5 figures

3D Gaussian Splats (3DGS) have proven a versatile rendering primitive, both for inverse rendering as well as real-time exploration of scenes. In these applications, coherence across camera frames and multiple views is crucial, be it for robust convergence of a scene reconstruction or for artifact-free fly-throughs. Recent work started mitigating artifacts that break multi-view coherence, including popping artifacts due to inconsistent transparency sorting and perspective-correct outlines of (2D) splats. At the same time, real-time requirements forced such implementations to accept compromises in how transparency of large assemblies of 3D Gaussians is resolved, in turn breaking coherence in other ways. In our work, we aim at achieving maximum coherence, by rendering fully perspective-correct 3D Gaussians while using a high-quality approximation of accurate blending, hybrid transparency, on a per-pixel level, in order to retain real-time frame rates. Our fast and perspectively accurate approach for evaluation of 3D Gaussians does not require matrix inversions, thereby ensuring numerical stability and eliminating the need for special handling of degenerate splats, and the hybrid transparency formulation for blending maintains similar quality as fully resolved per-pixel transparencies at a fraction of the rendering costs. We further show that each of these two components can be independently integrated into Gaussian splatting systems. In combination, they achieve up to 2$\times$ higher frame rates, 2$\times$ faster optimization, and equal or better image quality with fewer rendering artifacts compared to traditional 3DGS on common benchmarks., Comment: Project page: https://fhahlbohm.github.io/htgs/

The increasing size of deep learning models has created the need for more efficient alternatives to the standard error backpropagation algorithm, that make better use of asynchronous, parallel and distributed computing. One major shortcoming of backpropagation is the interlocking between the forward phase of the algorithm, which computes a global loss, and the backward phase where the loss is backpropagated through all layers to compute the gradients, which are used to update the network parameters. To address this problem, we propose a method that parallelises SGD updates across the layers of a model by asynchronously updating them from multiple threads. Furthermore, since we observe that the forward pass is often much faster than the backward pass, we use separate threads for the forward and backward pass calculations, which allows us to use a higher ratio of forward to backward threads than the usual 1:1 ratio, reducing the overall staleness of the parameters. Thus, our approach performs asynchronous stochastic gradient descent using separate threads for the loss (forward) and gradient (backward) computations and performs layer-wise partial updates to parameters in a distributed way. We show that this approach yields close to state-of-the-art results while running up to 2.97x faster than Hogwild! scaled on multiple devices (Locally-Partitioned-Asynchronous-Parallel SGD). We theoretically prove the convergence of the algorithm using a novel theoretical framework based on stochastic differential equations and the drift diffusion process, by modeling the asynchronous parameter updates as a stochastic process., Comment: 16 pages, 4 figures

Recent advances in AI have led to significant results in robotic learning, but skills like grasping remain partially solved. Many recent works exploit synthetic grasping datasets to learn to grasp unknown objects. However, those datasets were generated using simple grasp sampling methods using priors. Recently, Quality-Diversity (QD) algorithms have been proven to make grasp sampling significantly more efficient. In this work, we extend QDG-6DoF, a QD framework for generating object-centric grasps, to scale up the production of synthetic grasping datasets. We propose a data augmentation method that combines the transformation of object meshes with transfer learning from previous grasping repertoires. The conducted experiments show that this approach reduces the number of required evaluations per discovered robust grasp by up to 20%. We used this approach to generate QDGset, a dataset of 6DoF grasp poses that contains about 3.5 and 4.5 times more grasps and objects, respectively, than the previous state-of-the-art. Our method allows anyone to easily generate data, eventually contributing to a large-scale collaborative dataset of synthetic grasps., Comment: 8 pages, 9 figures. Draft version

Traditional nanophotonic sensing schemes utilize evanescent fields in dielectric or metallic nanoparticles, which confine far-field radiation in dispersive and lossy media. Apart from the lack of a well-defined sensing volume that can be accompanied by moderate sensitivities, these structures suffer from the generally limited access to the modal field, which is key for sensing performance. Recently, a novel strategy for dielectric nanophotonics has been demonstrated, namely, the resonant confinement of light in air. So-called Mie voids created in high-index dielectric host materials support localized resonant modes with exceptional properties. In particular, due to the confinement in air, these structures benefit from the full access to the modal field inside the void. We utilize these Mie voids for refractive index sensing in single voids with volumes down to 100 attoliters and sensitivities on the order of 400 nm per refractive index unit. Taking the signal-to-noise ratio of our measurements into account, we demonstrate detection of refractive index changes as small as 6.9 x 10-4 in a defined volume of just 850 attoliters. The combination of our Mie void sensor platform with appropriate surface functionalization will even enable specificity to biological or other analytes of interest, as the sensing volumes are on the order of cellular signaling chemicals of single vesicles in cellular synapses.

Dynamic reconstruction and spatiotemporal novel-view synthesis of non-rigidly deforming scenes recently gained increased attention. While existing work achieves impressive quality and performance on multi-view or teleporting camera setups, most methods fail to efficiently and faithfully recover motion and appearance from casual monocular captures. This paper contributes to the field by introducing a new method for dynamic novel view synthesis from monocular video, such as casual smartphone captures. Our approach represents the scene as a $\textit{dynamic neural point cloud}$, an implicit time-conditioned point distribution that encodes local geometry and appearance in separate hash-encoded neural feature grids for static and dynamic regions. By sampling a discrete point cloud from our model, we can efficiently render high-quality novel views using a fast differentiable rasterizer and neural rendering network. Similar to recent work, we leverage advances in neural scene analysis by incorporating data-driven priors like monocular depth estimation and object segmentation to resolve motion and depth ambiguities originating from the monocular captures. In addition to guiding the optimization process, we show that these priors can be exploited to explicitly initialize our scene representation to drastically improve optimization speed and final image quality. As evidenced by our experimental evaluation, our dynamic point cloud model not only enables fast optimization and real-time frame rates for interactive applications, but also achieves competitive image quality on monocular benchmark sequences. Our project page is available at https://moritzkappel.github.io/projects/dnpc., Comment: 16 pages, 5 figures, 10 tables. Project page: https://moritzkappel.github.io/projects/dnpc

Activity and parameter sparsity are two standard methods of making neural networks computationally more efficient. Event-based architectures such as spiking neural networks (SNNs) naturally exhibit activity sparsity, and many methods exist to sparsify their connectivity by pruning weights. While the effect of weight pruning on feed-forward SNNs has been previously studied for computer vision tasks, the effects of pruning for complex sequence tasks like language modeling are less well studied since SNNs have traditionally struggled to achieve meaningful performance on these tasks. Using a recently published SNN-like architecture that works well on small-scale language modeling, we study the effects of weight pruning when combined with activity sparsity. Specifically, we study the trade-off between the multiplicative efficiency gains the combination affords and its effect on task performance for language modeling. To dissect the effects of the two sparsities, we conduct a comparative analysis between densely activated models and sparsely activated event-based models across varying degrees of connectivity sparsity. We demonstrate that sparse activity and sparse connectivity complement each other without a proportional drop in task performance for an event-based neural network trained on the Penn Treebank and WikiText-2 language modeling datasets. Our results suggest sparsely connected event-based neural networks are promising candidates for effective and efficient sequence modeling., Comment: arXiv admin note: text overlap with arXiv:2311.07625

Event-based sensors are well suited for real-time processing due to their fast response times and encoding of the sensory data as successive temporal differences. These and other valuable properties, such as a high dynamic range, are suppressed when the data is converted to a frame-based format. However, most current methods either collapse events into frames or cannot scale up when processing the event data directly event-by-event. In this work, we address the key challenges of scaling up event-by-event modeling of the long event streams emitted by such sensors, which is a particularly relevant problem for neuromorphic computing. While prior methods can process up to a few thousand time steps, our model, based on modern recurrent deep state-space models, scales to event streams of millions of events for both training and inference. We leverage their stable parameterization for learning long-range dependencies, parallelizability along the sequence dimension, and their ability to integrate asynchronous events effectively to scale them up to long event streams. We further augment these with novel event-centric techniques enabling our model to match or beat the state-of-the-art performance on several event stream benchmarks. In the Spiking Speech Commands task, we improve state-of-the-art by a large margin of 7.7% to 88.4%. On the DVS128-Gestures dataset, we achieve competitive results without using frames or convolutional neural networks. Our work demonstrates, for the first time, that it is possible to use fully event-based processing with purely recurrent networks to achieve state-of-the-art task performance in several event-based benchmarks.

We introduce a new approach for reconstruction and novel-view synthesis of unbounded real-world scenes. In contrast to previous methods using either volumetric fields, grid-based models, or discrete point cloud proxies, we propose a hybrid scene representation, which implicitly encodes a point cloud in a continuous octree-based probability field and a multi-resolution hash grid. In doing so, we combine the benefits of both worlds by retaining favorable behavior during optimization: Our novel implicit point cloud representation and differentiable bilinear rasterizer enable fast rendering while preserving fine geometric detail without depending on initial priors like structure-from-motion point clouds. Our method achieves state-of-the-art image quality on several common benchmark datasets. Furthermore, we achieve fast inference at interactive frame rates, and can extract explicit point clouds to further enhance performance., Comment: Project page: https://fhahlbohm.github.io/inpc/

Recent advances in AI have led to significant results in robotic learning, including natural language-conditioned planning and efficient optimization of controllers using generative models. However, the interaction data remains the bottleneck for generalization. Getting data for grasping is a critical challenge, as this skill is required to complete many manipulation tasks. Quality-Diversity (QD) algorithms optimize a set of solutions to get diverse, high-performing solutions to a given problem. This paper investigates how QD can be combined with priors to speed up the generation of diverse grasps poses in simulation compared to standard 6-DoF grasp sampling schemes. Experiments conducted on 4 grippers with 2-to-5 fingers on standard objects show that QD outperforms commonly used methods by a large margin. Further experiments show that QD optimization automatically finds some efficient priors that are usually hard coded. The deployment of generated grasps on a 2-finger gripper and an Allegro hand shows that the diversity produced maintains sim-to-real transferability. We believe these results to be a significant step toward the generation of large datasets that can lead to robust and generalizing robotic grasping policies., Comment: 7 pages, 8 figures. Preprint version

As humans advance toward a higher level of artificial intelligence, it is always at the cost of escalating computational resource consumption, which requires developing novel solutions to meet the exponential growth of AI computing demand. Neuromorphic hardware takes inspiration from how the brain processes information and promises energy-efficient computing of AI workloads. Despite its potential, neuromorphic hardware has not found its way into commercial AI data centers. In this article, we try to analyze the underlying reasons for this and derive requirements and guidelines to promote neuromorphic systems for efficient and sustainable cloud computing: We first review currently available neuromorphic hardware systems and collect examples where neuromorphic solutions excel conventional AI processing on CPUs and GPUs. Next, we identify applications, models and algorithms which are commonly deployed in AI data centers as further directions for neuromorphic algorithms research. Last, we derive requirements and best practices for the hardware and software integration of neuromorphic systems into data centers. With this article, we hope to increase awareness of the challenges of integrating neuromorphic hardware into data centers and to guide the community to enable sustainable and energy-efficient AI at scale., Comment: 11 pages, 2 figures, presented as poster at NICE 2024, 2nd version with updated author list and minor updates

As large language models continue to scale in size rapidly, so too does the computational power required to run them. Event-based networks on neuromorphic devices offer a potential way to reduce energy consumption for inference significantly. However, to date, most event-based networks that can run on neuromorphic hardware, including spiking neural networks (SNNs), have not achieved task performance even on par with LSTM models for language modeling. As a result, language modeling on neuromorphic devices has seemed a distant prospect. In this work, we demonstrate the first-ever implementation of a language model on a neuromorphic device - specifically the SpiNNaker 2 chip - based on a recently published event-based architecture called the EGRU. SpiNNaker 2 is a many-core neuromorphic chip designed for large-scale asynchronous processing, while the EGRU is architected to leverage such hardware efficiently while maintaining competitive task performance. This implementation marks the first time a neuromorphic language model matches LSTMs, setting the stage for taking task performance to the level of large language models. We also demonstrate results on a gesture recognition task based on inputs from a DVS camera. Overall, our results showcase the feasibility of this neuro-inspired neural network in hardware, highlighting significant gains versus conventional hardware in energy efficiency for the common use case of single batch inference.

The nucleus of comet 67P/Churyumov-Gerasimenko exhibits a broad spectral reflectance feature around 3.2 $\mu$m, which is omnipresent in all spectra of the surface, and whose attribution has remained elusive since its discovery. Based on laboratory experiments, we have shown that most of this absorption feature is due to ammonium (NH4+) salts mixed with the dark surface material. The depth of the band is compatible with semi-volatile ammonium salts being a major reservoir of nitrogen in the comet, which could dominate over refractory organic matter and volatile species. These salts may thus represent the long-sought reservoir of nitrogen in comets, possibly bringing their nitrogen-to-carbon ratio in agreement with the solar value. Moreover, the reflectance spectra of several asteroids are compatible with the presence of NH4+ salts at their surfaces. The presence of such salts, and other NH4+-bearing compounds on asteroids, comets, and possibly in proto-stellar environments, suggests that NH4+ may be a tracer of the incorporation and transformation of nitrogen in ices, minerals and organics, at different phases of the formation of the Solar System.

The separation between the last closed flux surface of a plasma and the external coils that magnetically confine it is a limiting factor in the construction of fusion-capable plasma devices. This plasma-coil separation must be large enough so that components such as a breeding blanket and neutron shielding can fit between the plasma and the coils. Plasma-coil separation affects reactor size, engineering complexity, and particle loss due to field ripple. For some plasmas it can be difficult to produce the desired flux surface shaping with distant coils, and for other plasmas it is infeasible altogether. Here, we seek to understand the underlying physics that limits plasma-coil separation and explain why some configurations require close external coils. In this paper, we explore the hypothesis that the limiting plasma-coil separation is set by the shortest scale length of the magnetic field as expressed by the $\nabla$B tensor. We tested this hypothesis on a database of > 40 stellarator and tokamak configurations. Within this database, the coil-to-plasma distance compared to the minor radius varies by over an order of magnitude. The magnetic scale length is well correlated to the coil-to-plasma distance of actual coil designs generated using the REGCOIL method [Landreman, Nucl. Fusion 57, 046003 (2017)]. Additionally, this correlation reveals a general trend that larger plasma-coil separation is possible with a small number of field periods., Comment: 16 pages, 11 figures; fixed typos, added supplemental material Zenodo DOI

The perception of the value and propriety of modern engineered systems is changing. In addition to their functional and extra-functional properties, nowadays' systems are also evaluated by their sustainability properties. The next generation of systems will be characterized by an overall elevated sustainability -- including their post-life, driven by efficient value retention mechanisms. Current systems engineering practices fall short of supporting these ambitions and need to be revised appropriately. In this paper, we introduce the concept of circular systems engineering, a novel paradigm for systems sustainability, and define two principles to successfully implement it: end-to-end sustainability and bipartite sustainability. We outline typical organizational evolution patterns that lead to the implementation and adoption of circularity principles, and outline key challenges and research opportunities.

Different real-world cognitive tasks evolve on different relevant timescales. Processing these tasks requires memory mechanisms able to match their specific time constants. In particular, the working memory utilizes mechanisms that span orders of magnitudes of timescales, from milliseconds to seconds or even minutes. This plentitude of timescales is an essential ingredient of working memory tasks like visual or language processing. This degree of flexibility is challenging in analog computing hardware because it requires the integration of several reconfigurable capacitors of different size. Emerging volatile memristive devices present a compact and appealing solution to reproduce reconfigurable temporal dynamics in a neuromorphic network. We present a demonstration of working memory using a silver-based memristive device whose key parameters, retention time and switching probability, can be electrically tuned and adapted to the task at hand. First, we demonstrate the principles of working memory in a small scale hardware to execute an associative memory task. Then, we use the experimental data in two larger scale simulations, the first featuring working memory in a biological environment, the second demonstrating associative symbolic working memory.

The ubiquitous backpropagation algorithm requires sequential updates through the network introducing a locking problem. In addition, back-propagation relies on the transpose of forward weight matrices to compute updates, introducing a weight transport problem across the network. Locking and weight transport are problems because they prevent efficient parallelization and horizontal scaling of the training process. We propose a new method to address both these problems and scale up the training of large models. Our method works by dividing a deep neural network into blocks and introduces a feedback network that propagates the information from the targets backwards to provide auxiliary local losses. Forward and backward propagation can operate in parallel and with different sets of weights, addressing the problems of locking and weight transport. Our approach derives from a statistical interpretation of training that treats output activations of network blocks as parameters of probability distributions. The resulting learning framework uses these parameters to evaluate the agreement between forward and backward information. Error backpropagation is then performed locally within each block, leading to "block-local" learning. Several previously proposed alternatives to error backpropagation emerge as special cases of our model. We present results on a variety of tasks and architectures, demonstrating state-of-the-art performance using block-local learning. These results provide a new principled framework for training networks in a distributed setting., Comment: 23 pages, 4 figures, preprint

The reconstruction and novel view synthesis of dynamic scenes recently gained increased attention. As reconstruction from large-scale multi-view data involves immense memory and computational requirements, recent benchmark datasets provide collections of single monocular views per timestamp sampled from multiple (virtual) cameras. We refer to this form of inputs as "monocularized" data. Existing work shows impressive results for synthetic setups and forward-facing real-world data, but is often limited in the training speed and angular range for generating novel views. This paper addresses these limitations and proposes a new method for full 360{\deg} inward-facing novel view synthesis of non-rigidly deforming scenes. At the core of our method are: 1) An efficient deformation module that decouples the processing of spatial and temporal information for accelerated training and inference; and 2) A static module representing the canonical scene as a fast hash-encoded neural radiance field. In addition to existing synthetic monocularized data, we systematically analyze the performance on real-world inward-facing scenes using a newly recorded challenging dataset sampled from a synchronized large-scale multi-view rig. In both cases, our method is significantly faster than previous methods, converging in less than 7 minutes and achieving real-time framerates at 1K resolution, while obtaining a higher visual accuracy for generated novel views. Our source code and data is available at our project page https://graphics.tu-bs.de/publications/kappel2022fast., Comment: 18 pages, 14 figures; project page: https://graphics.tu-bs.de/publications/kappel2022fast