Data parallelism across multiple machines is widely adopted for accelerating distributed deep learning, but it is hard to achieve linear speedup due to the heavy communication. In this paper, we propose SAPipe, a performant system that pushes the training speed of data parallelism to its fullest extent.

Machine unlearning aims to remove the influence of specific data points from a trained model to satisfy privacy, copyright, and safety requirements. In real deployments, providers distribute a global model to many edge devices, where each client personalizes the model using private data. When a deletion request is issued, clients may ignore it or falsely claim compliance, and providers cannot check their parameters or data. This makes verification difficult, especially because personalized models must forget the targeted samples while preserving local utility, and verification must remain lightweight on edge devices. We introduce ZK APEX, a zero-shot personalized unlearning method that operates directly on the personalized model without retraining. ZK APEX combines sparse masking on the provider side with a small Group OBS compensation step on the client side, using a blockwise empirical Fisher matrix to create a curvature-aware update designed for low overhead. Paired with Halo2 zero-knowledge proofs, it enables the provider to verify that the correct unlearning transformation was applied without revealing any private data or personalized parameters. On Vision Transformer classification tasks, ZK APEX recovers nearly all personalization accuracy while effectively removing the targeted information. Applied to the OPT125M generative model trained on code data, it recovers around seventy percent of the original accuracy. Proof generation for the ViT case completes in about two hours, more than ten million times faster than retraining-based checks, with less than one gigabyte of memory use and proof sizes around four hundred megabytes. These results show the first practical framework for verifiable personalized unlearning on edge devices.

Conventional models of matching markets assume that monetary transfers can clear markets by compensating for utility differentials. However, empirical patterns show that such transfers often fail to close structural preference gaps. This paper introduces a market microstructure framework that models matching decisions as a limit order book system with rigid bid ask spreads. Individual preferences are represented by a latent preference state matrix, where the spread between an agent's internal ask price (the unconditional maximum) and the market's best bid (the reachable maximum) creates a structural liquidity constraint. We establish a Threshold Impossibility Theorem showing that linear compensation cannot close these spreads unless it induces a categorical identity shift. A dynamic discrete choice execution model further demonstrates that matches occur only when the market to book ratio crosses a time decaying liquidity threshold, analogous to order execution under inventory pressure. Numerical experiments validate persistent slippage, regional invariance of preference orderings, and high tier zero spread executions. The model provides a unified microstructure explanation for matching failures, compensation inefficiency, and post match regret in illiquid order driven environments.

In each step, one short-term player arrives at the system.

Evaluating large language models (LLMs) is challenging.

Our insight is that satellite's earth observation photos are not just images

Y et, we have a limited understanding of LLMs' communication and decision-making abilities in multi-agent setups.

Abstract--Autonomous robotic systems heavily rely on environment knowledge to safely navigate. For search & rescue, a flying robot requires robust real-time perception, enabled by complementary sensors. IMU data constrains acceleration and rotation, whereas LiDAR measures accurate distances around the robot. Our new scan window uses non-uniform temporal knot placement to ensure continuity over the whole trajectory without additional scan delay. Moreover, we accelerate essential covariance and GMM computations with Kronecker sums and products by a factor of 3.3. An unscented transform de-skews surfels, while a splitting into intra-scan segments facilitates motion compensation during spline optimization. Complementary soft constraints on relative poses and preintegrated IMU pseudo-measurements further improve robustness and accuracy. ELIABLE real-time perception is essential for robotic autonomy. In particular, accurate mapping and ego-motion estimation are key components for safe interaction in complex and unstructured environments. Due to their precision and measurement density, modern LiDARs are often used in these scenarios, e.g., in the DARP A Subterranean Challenge [1], [2]. Sensor motion during scanning distorts the point cloud and degrades the quality of the map. This intra-scan motion is either compensated by de-skewing prior to registration [3], [4], [5], [6] or by modeling it with a continuous-time trajectory [7], [8], [9]. The former uses the previous state estimate and, optionally, an IMU to predict the motion and transform points to a common reference time. However, this comes at the cost of reduced real-time capability and requires either costly reintegration of surfels [9] or a limited number of selected pointwise features [e.g., CT -ICP [7], CLINS [8]]. To overcome these limitations of continuous-time methods, our novel real-time LiDAR-inertial odometry (LIO) jointly optimizes temporally partitioned scan segments (Figure 1) by registering multi-resolution surfel maps while an unscented transform (UT) compensates the intra-surfel motion. Manuscript received October XX, 2025; revised XX, 2025.

In each step, one short-term player arrives at the system.