We introduce a training-efficient framework for time-series learning that combines random features with controlled differential equations (CDEs). In this approach, large randomly parameterized CDEs act as continuous-time reservoirs, mapping input paths to rich representations. Only a linear readout layer is trained, resulting in fast, scalable models with strong inductive bias. Building on this foundation, we propose two variants: (i) Random Fourier CDEs (RF-CDEs): these lift the input signal using random Fourier features prior to the dynamics, providing a kernel-free approximation of RBF-enhanced sequence models; (ii) Random Rough DEs (R-RDEs): these operate directly on rough-path inputs via a log-ODE discretization, using log-signatures to capture higher-order temporal interactions while remaining stable and efficient. We prove that in the infinite-width limit, these model induces the RBF-lifted signature kernel and the rough signature kernel, respectively, offering a unified perspective on random-feature reservoirs, continuous-time deep architectures, and path-signature theory. We evaluate both models across a range of time-series benchmarks, demonstrating competitive or state-of-the-art performance. These methods provide a practical alternative to explicit signature computations, retaining their inductive bias while benefiting from the efficiency of random features.

As the world looks to incorporate more renewables into energy grids, centuries-old systems that can balance supply and demand are being reappraised and innovated upon.

Developmental Graph Cellular Automata (DGCA) are a novel model for morphogenesis, capable of growing directed graphs from single-node seeds. In this paper, we show that DGCAs can be trained to grow reservoirs. Reservoirs are grown with two types of targets: task-driven (using the NARMA family of tasks) and task-independent (using reservoir metrics). Results show that DGCAs are able to grow into a variety of specialized, life-like structures capable of effectively solving benchmark tasks, statistically outperforming `typical' reservoirs on the same task. Overall, these lay the foundation for the development of DGCA systems that produce plastic reservoirs and for modeling functional, adaptive morphogenesis.

We present a novel approach to EEG decoding for non-invasive brain-machine interfaces (BMIs), with a focus on motor-behavior classification. While conventional con-volutional architectures such as EEGNet and DeepConvNet are effective in capturing local spatial patterns, they are markedly less suited for modeling long-range temporal dependencies and nonlinear dynamics [1, 2]. To address this limitation, we integrate an Echo State Network (ESN), a prominent paradigm in reservoir computing into the decoding pipeline [3, 4, 5]. ESNs construct a high-dimensional, sparsely connected recurrent reservoir that excels at tracking temporal dynamics, thereby complementing the spatial representational power of CNNs. Evaluated on a skateboard-trick EEG dataset preprocessed via the PREP pipeline and implemented in MNE-Python, our ESNNet achieves 83.2% within-subject and 51.3% LOSO accuracies, surpassing widely used CNN-based baselines [6, 7].

Because of shifting storms and sweltering summers, Iran's capital faces a future "Day Zero" when the taps run dry. During the summer of 2025, Iran experienced an exceptional heat wave, with daytime temperatures across several regions, including Tehran, approaching 50 degrees Celsius (122 degrees Fahrenheit) and forcing the temporary closure of public offices and banks. During this period, major reservoirs supplying the Tehran region reached record-low levels, and water supply systems came under acute strain . By early November, the reservoir behind Amir Kabir Dam, a main source of drinking water for Tehran, had dropped to about 8 percent of its capacity . The present crisis reflects not only this summer's extreme heat but also several consecutive years of reduced precipitation and ongoing drought conditions across Iran.

RMSProp avoids both the vanishing and exploding gradient problems.

We thank the reviewers for clear and thoughtful feedback, and respond to specific points raised by reviewers below. This comparison ([22]) is representative of "train[ing] an agent and task Our approach outperforms [22] on transfer to test tasks. R3: "Show that ... the newly proposed task is super useful". We agree with R2 that more sophisticated sampling strategies are worth pursuing in future work. We use the same hyper-parameters for skill acquisition (i.e.

Automatic cocktail machines are silly, but also kind of fun. Here's how to choose between the Bartesian and Barsys devices. The machine on my kitchen table is a holy device, if your definition of "holy" is that it looks like a glowing halo and it's filled with spirits. The machine has taken up a task I consider sacred: making me a cocktail. In advance of holiday party season, I have been testing a pair of devices that promise an indulgent future, a life where machines can make you a passable Old Fashioned.

Cortical networks are strongly recurrent, and neurons have intrinsic temporal dynamics.