A critical barrier to the trustworthiness of sixth-generation (6G) agentic autonomous networks is the uncertainty neglect bias; a cognitive tendency for large language model (LLM)-powered agents to make high-stakes decisions based on simple averages while ignoring the tail risk of extreme events. This paper proposes an unbiased, risk-aware framework for agentic negotiation, designed to ensure robust resource allocation in 6G network slicing. Specifically, agents leverage Digital Twins (DTs) to predict full latency distributions, which are then evaluated using a formal framework from extreme value theory, namely, Conditional Value-at-Risk (CVaR). This approach fundamentally shifts the agent's objective from reasoning over the mean to reasoning over the tail, thereby building a statistically-grounded buffer against worst-case outcomes. Furthermore, our framework ensures full uncertainty awareness by requiring agents to quantify epistemic uncertainty -- confidence in their own DTs predictions -- and propagate this meta-verification to make robust decisions, preventing them from acting on unreliable data. We validate this framework in a 6G inter-slice negotiation use-case between an eMBB and a URLLC agent. The results demonstrate the profound failure of the biased, mean-based baseline, which consistently fails its SLAs with a 25\% rate. Our unbiased, CVaR-aware agent successfully mitigates this bias, eliminating SLA violations and reducing the URLLC and eMBB p99.999 latencies by around 11\%. We show this reliability comes at the rational and quantifiable cost of slightly reduced energy savings to 17\%, exposing the false economy of the biased approach. This work provides a concrete methodology for building the trustworthy autonomous systems required for 6G.

In Diffusion Transformer (DiT) models, particularly for video generation, attention latency is a major bottleneck due to the long sequence length and the quadratic complexity. We find that attention weights can be separated into two parts: a small fraction of large weights with high rank and the remaining weights with very low rank. This naturally suggests applying sparse acceleration to the first part and low-rank acceleration to the second. Based on this finding, we propose SLA (Sparse-Linear Attention), a trainable attention method that fuses sparse and linear attention to accelerate diffusion models. SLA classifies attention weights into critical, marginal, and negligible categories, applying O(N^2) attention to critical weights, O(N) attention to marginal weights, and skipping negligible ones. SLA combines these computations into a single GPU kernel and supports both forward and backward passes. With only a few fine-tuning steps using SLA, DiT models achieve a 20x reduction in attention computation, resulting in significant acceleration without loss of generation quality. Experiments show that SLA reduces attention computation by 95% without degrading end-to-end generation quality, outperforming baseline methods. In addition, we implement an efficient GPU kernel for SLA, which yields a 13.7x speedup in attention computation and a 2.2x end-to-end speedup in video generation on Wan2.1-1.3B. The code is available at https://github.com/thu-ml/SLA.

Cloud computing has established itself as the support for the vast majority of emerging technologies, mainly due to the characteristic of elasticity it offers. Auto-scalers are the systems that enable this elasticity by acquiring and releasing resources on demand to ensure an agreed service level. In this article we present FLAS (Forecasted Load Auto-Scaling), an auto-scaler for distributed services that combines the advantages of proactive and reactive approaches according to the situation to decide the optimal scaling actions in every moment. The main novelties introduced by FLAS are (i) a predictive model of the high-level metrics trend which allows to anticipate changes in the relevant SLA parameters (e.g. performance metrics such as response time or throughput) and (ii) a reactive contingency system based on the estimation of high-level metrics from resource use metrics, reducing the necessary instrumentation (less invasive) and allowing it to be adapted agnostically to different applications. We provide a FLAS implementation for the use case of a content-based publish-subscribe middleware (E-SilboPS) that is the cornerstone of an event-driven architecture. To the best of our knowledge, this is the first auto-scaling system for content-based publish-subscribe distributed systems (although it is generic enough to fit any distributed service). Through an evaluation based on several test cases recreating not only the expected contexts of use, but also the worst possible scenarios (following the Boundary-Value Analysis or BVA test methodology), we have validated our approach and demonstrated the effectiveness of our solution by ensuring compliance with performance requirements over 99% of the time.

Neural Information Processing Systems http://nips.cc/

Autoregressive decoding algorithms that use only past information often cannot guarantee the best performance. Recently, people discovered that looking-ahead algorithms such as Monte Carlo Tree Search (MCTS) with external reward models (RMs) can significantly improve models' output by allowing them to think ahead and leverage future outputs and associated rewards to guide the current generation. Such techniques can help the reinforcement fine-tuning phase by sampling better trajectories and the inference phase by selecting the better output. However, their high computational cost limits their applications, especially in streaming scenarios. To address this issue, we propose equipping the policy model with token-level self-reward modeling (TRM) capability to eliminate the need for external models and extra communication. We name the new architecture as Reward Transformer. In addition, we propose a streaming-looking-ahead (SLA) algorithm to further boost search efficiency with better parallelization. Experiments show that SLA achieves an overall win rate of 79.7\% against the baseline greedy decoding algorithm on three general-domain datasets with a frozen policy model while maintaining streaming efficiency. If we combine SLA with reinforcement fine-tuning techniques such as DPO, SLA achieves an overall win rate of 89.4\%.

Single-snapshot signal processing in sparse linear arrays has become increasingly vital, particularly in dynamic environments like automotive radar systems, where only limited snapshots are available. These arrays are often utilized either to cut manufacturing costs or result from unintended antenna failures, leading to challenges such as high sidelobe levels and compromised accuracy in direction-of-arrival (DOA) estimation. Despite deep learning's success in tasks such as DOA estimation, the need for extensive training data to increase target numbers or improve angular resolution poses significant challenges. In response, this paper presents a novel Siamese neural network (SNN) featuring a sparse augmentation layer, which enhances signal feature embedding and DOA estimation accuracy in sparse arrays. We demonstrate the enhanced DOA estimation performance of our approach through detailed feature analysis and performance evaluation. The code for this study is available at https://github.com/ruxinzh/SNNS_SLA.

Open-weight large language model (LLM) zoos allow users to quickly integrate state-of-the-art models into systems. Despite increasing availability, selecting the most appropriate model for a given task still largely relies on public benchmark leaderboards and educated guesses. This can be unsatisfactory for both inference service providers and end users, where the providers usually prioritize cost efficiency, while the end users usually prioritize model output quality for their inference requests. In commercial settings, these two priorities are often brought together in Service Level Agreements (SLA). We present MESS+, an online stochastic optimization algorithm for energy-optimal model selection from a model zoo, which works on a per-inference-request basis. For a given SLA that requires high accuracy, we are up to 2.5x more energy efficient with MESS+ than with randomly selecting an LLM from the zoo while maintaining SLA quality constraints.

Speech is a fundamental aspect of human life, crucial not only for communication but also for cognitive, social, and academic development. Children with speech disorders (SD) face significant challenges that, if unaddressed, can result in lasting negative impacts. Traditionally, speech and language assessments (SLA) have been conducted by skilled speech-language pathologists (SLPs), but there is a growing need for efficient and scalable SLA methods powered by artificial intelligence. This position paper presents a survey of existing techniques suitable for automating SLA pipelines, with an emphasis on adapting automatic speech recognition (ASR) models for children's speech, an overview of current SLAs and their automated counterparts to demonstrate the feasibility of AI-enhanced SLA pipelines, and a discussion of practical considerations, including accessibility and privacy concerns, associated with the deployment of AI-powered SLAs.

This paper introduces a novel framework combining LLM agents as proxies for human strategic behavior with reinforcement learning (RL) to engage these agents in evolving strategic interactions within team environments. Our approach extends traditional agent-based simulations by using strategic LLM agents (SLA) and introducing dynamic and adaptive governance through a pro-social promoting RL agent (PPA) that modulates information access across agents in a network, optimizing social welfare and promoting pro-social behavior. Through validation in iterative games, including the prisoner's dilemma, we demonstrate that SLA agents exhibit nuanced strategic adaptations. The PPA agent effectively learns to adjust information transparency, resulting in enhanced cooperation rates. This framework offers significant insights into AI-mediated social dynamics, contributing to the deployment of AI in real-world team settings.