Zerong Xi

Luoxi Tang, Rupali Rajendra Vaje, Yuqiao Meng et al.

Agentic memory enables LLMs to persist information beyond a single context window and reuse it in later decisions, but it also introduces a new vulnerability: spurious correlations, where retrieved memory carries miscorrelated evidence and propagates erroneous reasoning into downstream decisions. Despite the widespread use of agentic memory, this risk remains largely underexplored. We address it from two aspects. First, we benchmark several canonical types of spurious patterns identified through causal structure and record them across trajectory-level memory. Diagnosing agentic memory systems on this benchmark reveals that memory improves reasoning on clean inputs but amplifies reliance on spurious patterns when they are present. Second, we propose CAMEL, a plug-and-play calibration method that operates across diverse memory architectures at both write and retrieval time. CAMEL consistently reduces reliance on spurious patterns across all three types while preserving or improving performance on clean inputs and staying robust under adaptive attacks targeting the calibration. Overall, CAMEL offers a principled and lightweight solution toward more reliable agentic memory deployment.

Luoxi Tang, Yuqiao Meng, Joseph Costa et al.

Multi-agent debate (MAD) systems improve LLM reasoning through iterative deliberation, but remain vulnerable to debate collapse, a failure type where final agent decisions are compromised on erroneous reasoning. Existing methods lack principled mechanisms to detect or prevent such failures. To address this gap, we first propose a hierarchical metric that quantifies behavioral uncertainty at three levels: intra-agent (individual reasoning uncertainty), inter-agent (interactive uncertainty), and system-level (output uncertainty). Empirical analysis across several benchmarks reveals that our proposed uncertainty quantification reliably indicates system failures, which demonstrates the validity of using them as diagnostic metrics to indicate the system failure. Subsequently, we propose a mitigation strategy by formulating an uncertainty-driven policy optimization to penalize self-contradiction, peer conflict, and low-confidence outputs in a dynamic debating environment. Experiments demonstrate that our proposed uncertainty-driven mitigation reliably calibrates the multi-agent system by consistently improving decision accuracy while reducing system disagreement.

Zerong Xi, Gita Sukthankar

This paper introduces a method for constructing an upper bound for exploration policy using either the weighted variance of return sequences or the weighted temporal difference (TD) error. We demonstrate that the variance of the return sequence for a specific state-action pair is an important information source that can be leveraged to guide exploration in reinforcement learning. The intuition is that fluctuation in the return sequence indicates greater uncertainty in the near future returns. This divergence occurs because of the cyclic nature of value-based reinforcement learning; the evolving value function begets policy improvements which in turn modify the value function. Although both variance and TD errors capture different aspects of this uncertainty, our analysis shows that both can be valuable to guide exploration. We propose a two-stream network architecture to estimate weighted variance/TD errors within DQN agents for our exploration method and show that it outperforms the baseline on a wide range of Atari games.

Yang Zhang, Zixiang Zhou, Philip David et al.

The need for fine-grained perception in autonomous driving systems has resulted in recently increased research on online semantic segmentation of single-scan LiDAR. Despite the emerging datasets and technological advancements, it remains challenging due to three reasons: (1) the need for near-real-time latency with limited hardware; (2) uneven or even long-tailed distribution of LiDAR points across space; and (3) an increasing number of extremely fine-grained semantic classes. In an attempt to jointly tackle all the aforementioned challenges, we propose a new LiDAR-specific, nearest-neighbor-free segmentation algorithm - PolarNet. Instead of using common spherical or bird's-eye-view projection, our polar bird's-eye-view representation balances the points across grid cells in a polar coordinate system, indirectly aligning a segmentation network's attention with the long-tailed distribution of the points along the radial axis. We find that our encoding scheme greatly increases the mIoU in three drastically different segmentation datasets of real urban LiDAR single scans while retaining near real-time throughput.

Jiaping Zhao, Zerong Xi, Laurent Itti

We propose to learn multiple local Mahalanobis distance metrics to perform k-nearest neighbor (kNN) classification of temporal sequences. Temporal sequences are first aligned by dynamic time warping (DTW); given the alignment path, similarity between two sequences is measured by the DTW distance, which is computed as the accumulated distance between matched temporal point pairs along the alignment path. Traditionally, Euclidean metric is used for distance computation between matched pairs, which ignores the data regularities and might not be optimal for applications at hand. Here we propose to learn multiple Mahalanobis metrics, such that DTW distance becomes the sum of Mahalanobis distances. We adapt the large margin nearest neighbor (LMNN) framework to our case, and formulate multiple metric learning as a linear programming problem. Extensive sequence classification results show that our proposed multiple metrics learning approach is effective, insensitive to the preceding alignment qualities, and reaches the state-of-the-art performances on UCR time series datasets.