Babak's Daily Briefing

Tuesday, April 7, 2026

Sources: 20 | Total Articles: 96

• 1.SoftBank’s Physical AI push gives AI-RAN a sharper purpose

  SoftBank is starting to give AI-RAN a more concrete job description: not just running AI workloads near the network, but serving as the real-time infrastructure layer for robots and other physical systems. The company’s recent materials suggest it wants to move the AI-RAN conversation from telecom architecture to real-world machine action.

• 2.South Korea puts 6G inside its national AI push

  South Korea has unveiled a three-year national roadmap aimed at becoming one of the world’s top three AI powers by 2028, with 6G commercialization positioned as part of that broader push.

• 3.b-com’s Open XG Hub targets one of telecom’s biggest gaps: turning experimentation into deployment

  In an interview with Peter Pietrzyk, Managing Director of 6GWorld, Patrick Savell, Head of Connectivity at b-com, said platforms such as Open XG Hub are designed to help bridge one of the industry’s most persistent challenges: moving promising ideas from research environments into deployable network systems. The bigger point is that, as telecom becomes more software-driven and AI-native, the bottleneck is increasingly less about invention and more about validation, integration, and operational readiness.

• 4.ODC’s $45M raise signals a bigger shift in AI-RAN, from network optimization to edge intelligence

  ORAN Development Company said it has closed a $45 million Series A backed by Booz Allen, Cisco Investments, Nokia, NVIDIA, AT&T, MTN and Telecom Italia to scale its U.S.-based Odyssey platform, which it positions as an AI-native RAN architecture combining communications, sensing and edge intelligence. The company said it plans to accelerate commercial deployment through 2026.

• 5.Lockheed Martin’s NetSense points to a bigger shift: 5G as drone-detection infrastructure

  Lockheed Martin’s latest NetSense prototype suggests that commercial 5G infrastructure could play a growing role in drone detection, adding momentum to the broader move toward sensing-enabled wireless networks.

• 1.Cheap Talk, Empty Promise: Frontier LLMs easily break public promises for self-interest

  Large language models are increasingly deployed as autonomous agents in multi-agent settings where they communicate intentions and take consequential actions with limited human oversight. A critical safety question is whether agents that publicly commit to actions break those promises when they can privately deviate, and what the consequences are for both themselves and the collective. We study deception as a deviation from a publicly announced action in one-shot normal-form games, classifying each deviation by its effect on individual payoff and collective welfare into four categories: win-win, selfish, altruistic, and sabotaging. By exhaustively enumerating announcement profiles across six canonical games, nine frontier models, and varying group sizes, we identify all opportunities for each deviation type and measure how often agents ex...

• 2.Springdrift: An Auditable Persistent Runtime for LLM Agents with Case-Based Memory, Normative Safety, and Ambient Self-Perception

  We present Springdrift, a persistent runtime for long-lived LLM agents. The system integrates an auditable execution substrate (append-only memory, supervised processes, git-backed recovery), a case-based reasoning memory layer with hybrid retrieval (evaluated against a dense cosine baseline), a deterministic normative calculus for safety gating with auditable axiom trails, and continuous ambient self-perception via a structured self-state representation (the sensorium) injected each cycle without tool calls. These properties support behaviours difficult to achieve in session-bounded systems: cross-session task continuity, cross-channel context maintenance, end-to-end forensic reconstruction of decisions, and self-diagnostic behaviour. We report on a single-instance deployment over 23 days (19 operating days), during which the agent diagn...

• 3.LOCARD: An Agentic Framework for Blockchain Forensics

  Blockchain forensics inherently involves dynamic and iterative investigations, while many existing approaches primarily model it through static inference pipelines. We propose a paradigm shift towards Agentic Blockchain Forensics (ABF), modeling forensic investigation as a sequential decision-making process. To instantiate this paradigm, we introduce LOCARD, the first agentic framework for blockchain forensics. LOCARD operationalizes this perspective through a Tri-Core Cognitive Architecture that decouples strategic planning, operational execution, and evaluative validation. Unlike generic LLM-based agents, it incorporates a Structured Belief State mechanism to enforce forensic rigor and guide exploration under explicit state constraints. To demonstrate the efficacy of the ABF paradigm, we apply LOCARD to the inherently complex domain of ...

• 4.Symbolic-Vector Attention Fusion for Collective Intelligence

  When autonomous agents observe different domains of a shared environment, each signal they exchange mixes relevant and irrelevant dimensions. No existing mechanism lets the receiver evaluate which dimensions to absorb. We introduce Symbolic-Vector Attention Fusion (SVAF), the content-evaluation half of a two-level coupling engine for collective intelligence. SVAF decomposes each inter-agent signal into 7 typed semantic fields, evaluates each through a learned fusion gate, and produces a remix -- new knowledge from the intersection of two domains. A band-pass model yields four outcomes (redundant, aligned, guarded, rejected), solving both selectivity and redundancy. The fusion gate independently discovers a cross-domain relevance hierarchy: mood emerges as the highest-weight field by epoch 1, before accuracy plateaus -- consistent with ind...

• 5.Your Agent is More Brittle Than You Think: Uncovering Indirect Injection Vulnerabilities in Agentic LLMs

  The rapid deployment of open-source frameworks has significantly advanced the development of modern multi-agent systems. However, expanded action spaces, including uncontrolled privilege exposure and hidden inter-system interactions, pose severe security challenges. Specifically, Indirect Prompt Injections (IPI), which conceal malicious instructions within third-party content, can trigger unauthorized actions such as data exfiltration during normal operations. While current security evaluations predominantly rely on isolated single-turn benchmarks, the systemic vulnerabilities of these agents within complex dynamic environments remain critically underexplored. To bridge this gap, we systematically evaluate six defense strategies against four sophisticated IPI attack vectors across nine LLM backbones. Crucially, we conduct our evaluation e...

• 1.Self-Execution Simulation Improves Coding Models

  arXiv:2604.03253v1 Announce Type: new Abstract: A promising research direction in enabling LLMs to generate consistently correct code involves addressing their inability to properly estimate program execution, particularly for code they generate. In this work, we demonstrate that Code LLMs can be trained to simulate program execution in a step-by-step manner and that this capability can be leveraged to improve competitive programming performance. Our approach combines supervised fine-tuning on natural language execution traces, textual explanations grounded in true execution, with reinforcement learning using verifiable rewards. We introduce two complementary objectives: output prediction given code and inputs, and solving competitive programming tasks with either ground-truth or self-predicted execution feedback. These objectives enable...

• 2.Robust LLM Performance Certification via Constrained Maximum Likelihood Estimation

  arXiv:2604.03257v1 Announce Type: new Abstract: The ability to rigorously estimate the failure rates of large language models (LLMs) is a prerequisite for their safe deployment. Currently, however, practitioners often face a tradeoff between expensive human gold standards and potentially severely-biased automatic annotation schemes such as "LLM-as-a-Judge" labeling. In this paper, we propose a new, practical, and efficient approach to LLM failure rate estimation based on constrained maximum-likelihood estimation (MLE). Our method integrates three distinct signal sources: (i) a small, high-quality human-labeled calibration set, (ii) a large corpus of LLM-judge annotations, and, most importantly, (iii) additional side information via domain-specific constraints derived from known bounds on judge performance statistics. We validate our appr...

• 3.SoLA: Leveraging Soft Activation Sparsity and Low-Rank Decomposition for Large Language Model Compression

  arXiv:2604.03258v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated impressive capabilities across various tasks, but the billion-scale parameters pose deployment challenges. Although existing methods attempt to reduce the scale of LLMs, they require either special hardware support or expensive post-training to maintain model quality. To facilitate efficient and affordable model slimming, we propose a novel training-free compression method for LLMs, named "SoLA", which leverages \textbf{So}ft activation sparsity and \textbf{L}ow-r\textbf{A}nk decomposition. SoLA can identify and retain a minority of components significantly contributing to inference, while compressing the majority through low-rank decomposition, based on our analysis of the activation pattern in the feed-forward network (FFN) of modern LLMs. To...

• 4.Why Attend to Everything? Focus is the Key

  arXiv:2604.03260v1 Announce Type: new Abstract: We introduce Focus, a method that learns which token pairs matter rather than approximating all of them. Learnable centroids assign tokens to groups; distant attention is restricted to same-group pairs while local attention operates at full resolution. Because all model weights stay frozen, Focus is purely additive: centroid-only training (as few as 148K parameters) improves domain perplexity with zero degradation on downstream benchmarks--from 124M to 70B parameters, across five attention architectures. No existing efficient attention method achieves this in the retrofit setting. At 124M, Focus surpasses full attention (30.3 vs 31.4 PPL); trained from scratch at 7B scale (2B tokens), Focus again beats full attention (13.82 vs 13.89 PPL). At inference, restricting each token to its top-k hi...

• 5.VIGIL: An Extensible System for Real-Time Detection and Mitigation of Cognitive Bias Triggers

  arXiv:2604.03261v1 Announce Type: new Abstract: The rise of generative AI is posing increasing risks to online information integrity and civic discourse. Most concretely, such risks can materialise in the form of mis- and disinformation. As a mitigation, media-literacy and transparency tools have been developed to address factuality of information and the reliability and ideological leaning of information sources. However, a subtler but possibly no less harmful threat to civic discourse is to use of persuasion or manipulation by exploiting human cognitive biases and related cognitive limitations. To the best of our knowledge, no tools exist to directly detect and mitigate the presence of triggers of such cognitive biases in online information. We present VIGIL (VIrtual GuardIan angeL), the first browser extension for real-time cognitive ...

• 1.Convolutional Surrogate for 3D Discrete Fracture-Matrix Tensor Upscaling

  arXiv:2604.02335v1 Announce Type: new Abstract: Modeling groundwater flow in three-dimensional fractured crystalline media requires accounting for strong spatial heterogeneity induced by fractures. Fine-scale discrete fracture-matrix (DFM) simulations can capture this complexity but are computationally expensive, especially when repeated evaluations are needed. To address this, we aim to employ a multilevel Monte Carlo (MLMC) framework in which numerical homogenization is used to upscale sub-resolution fracture effects when transitioning between accuracy levels. To reduce the cost of conventional 3D numerical homogenization, we develop a surrogate model that predicts the equivalent hydraulic conductivity tensor Keq from a voxelized 3D domain representing tensor-valued random fields of matrix and fracture conductivities. Fracture size, ori...

• 2.Generating Counterfactual Patient Timelines from Real-World Data

  arXiv:2604.02337v1 Announce Type: new Abstract: Counterfactual simulation - exploring hypothetical consequences under alternative clinical scenarios - holds promise for transformative applications such as personalized medicine and in silico trials. However, it remains challenging due to methodological limitations. Here, we show that an autoregressive generative model trained on real-world data from over 300,000 patients and 400 million patient timeline entries can generate clinically plausible counterfactual trajectories. As a validation task, we applied the model to patients hospitalized with COVID-19 in 2023, modifying age, serum C-reactive protein (CRP), and serum creatinine to simulate 7-day outcomes. Increased in-hospital mortality was observed in counterfactual simulations with older age, elevated CRP, and elevated serum creatinine....

• 3.LiME: Lightweight Mixture of Experts for Efficient Multimodal Multi-task Learning

  arXiv:2604.02338v1 Announce Type: new Abstract: MoE-PEFT methods combine Mixture of Experts with parameter-efficient fine-tuning for multi-task adaptation, but require separate adapters per expert causing trainable parameters to scale linearly with expert count and limiting applicability to adapter-based architectures. We propose LiME (Lightweight Mixture of Experts), which achieves expert specialization through lightweight modulation rather than adapter replication. Instead of separate adapters, LiME uses a single shared PEFT module and modulates its output with lightweight expert vectors, reducing expert parameters while generalizing to any PEFT method. Notably, LiME introduces zero-parameter routing by leveraging existing frozen and adapted representations eliminating learned router parameters typically required per layer. Theoreticall...

• 4.SIEVE: Sample-Efficient Parametric Learning from Natural Language

  arXiv:2604.02339v1 Announce Type: new Abstract: Natural language context-such as instructions, knowledge, or feedback-contains rich signal for adapting language models. While in-context learning provides adaptation via the prompt, parametric learning persists into model weights and can improve performance further, though is data hungry and heavily relies on either high-quality traces or automated verifiers. We propose SIEVE, a method for sample-efficient parametric learning from natural language context that requires as few as three query examples. SIEVE uses a novel synthetic data generation pipeline, SIEVE-GEN, that leverages the insight that context is decomposable. Decomposing context allows us to generate higher quality rollouts by pairing synthetic queries with only the applicable context rather than the entirety, then using context...

• 5.Not All Denoising Steps Are Equal: Model Scheduling for Faster Masked Diffusion Language Models

  arXiv:2604.02340v1 Announce Type: new Abstract: Recent advances in masked diffusion language models (MDLMs) narrow the quality gap to autoregressive LMs, but their sampling remains expensive because generation requires many full-sequence denoising passes with a large Transformer and, unlike autoregressive decoding, cannot benefit from KV caching. In this work, we exploit the flexibility of the diffusion framework and study model scheduling, where a smaller MDLM replaces the full model at a subset of denoising steps. On OpenWebText, we show that early and late denoising steps are substantially more robust to such replacement than middle steps, enabling up to a 17% reduction in FLOPs with only modest degradation in generative perplexity. We support these findings with a step-importance analysis based on loss and KL divergence between small ...

• 1.Surrogate Model-Based Near-Optimal Gain Selection for Approach-Angle-Constrained Two-Phase Pure Proportional Navigation

  arXiv:2604.03371v1 Announce Type: new Abstract: In guidance literature, Pure Proportional Navigation (PPN) guidance is widely used for aerodynamically driven vehicles. A two-phase extension of PPN (2pPPN), which uses different navigation gains for an orientation phase and a final phase, has been presented to achieve any desired approach angle within an angular half-space. Recent studies show that the orientation phase can be realized through multiple feasible trajectories, creating an opportunity to select navigation gains that minimize overall guidance effort. This paper addresses the problem of near-optimal gain selection for given initial and desired terminal engagement geometries. Two optimization problems are considered: i) determination of the optimal orientation-phase gain for a specified final-phase gain, and ii) simultaneously de...

• 2.Activity-Dependent Plasticity in Morphogenetically-Grown Recurrent Networks

  arXiv:2604.03386v1 Announce Type: new Abstract: Developmental approaches to neural architecture search grow functional networks from compact genomes through self-organisation, but the resulting networks operate with fixed post-growth weights. We characterise Hebbian and anti-Hebbian plasticity across 50,000 morphogenetically grown recurrent controllers (5M+ configurations on CartPole and Acrobot), then test whether co-evolutionary experiments -- where plasticity parameters are encoded in the genome and evolved alongside the developmental architecture -- recover these patterns independently. Our characterisation reveals that (1) anti-Hebbian plasticity significantly outperforms Hebbian for competent networks (Cohen's d = 0.53-0.64), (2) regret (fraction of oracle improvement lost under the best fixed setting) reaches 52-100%, and (3) plast...

• 3.Learning-Based Fault Detection for Legged Robots in Remote Dynamic Environments

  arXiv:2604.03397v1 Announce Type: new Abstract: Operations in hazardous environments put humans, animals, and machines at high risk for physically damaging consequences. In contrast to humans and animals, quadruped robots cannot naturally identify and adjust their locomotion to a severely debilitated limb. The ability to detect limb damage and adjust movement to a new physical morphology is the difference between survival and death for humans and animals. The same can be said for quadruped robots autonomously carrying out remote assignments in dynamic, complex settings. This work presents the development and implementation of an off-line learning-based method to detect single limb faults from proprioceptive sensor data in a quadrupedal robot. The aim of the fault detection technique is to provide the correct output for the controller to s...

• 4.Diffusion Policy with Bayesian Expert Selection for Active Multi-Target Tracking

  arXiv:2604.03404v1 Announce Type: new Abstract: Active multi-target tracking requires a mobile robot to balance exploration for undetected targets with exploitation of uncertain tracked ones. Diffusion policies have emerged as a powerful approach for capturing diverse behavioral strategies by learning action sequences from expert demonstrations. However, existing methods implicitly select among strategies through the denoising process, without uncertainty quantification over which strategy to execute. We formulate expert selection for diffusion policies as an offline contextual bandit problem and propose a Bayesian framework for pessimistic, uncertainty-aware strategy selection. A multi-head Variational Bayesian Last Layer (VBLL) model predicts the expected tracking performance of each expert strategy given the current belief state, provi...

• 5.Do Robots Need Body Language? Comparing Communication Modalities for Legible Motion Intent in Human-Shared Spaces

  arXiv:2604.03451v1 Announce Type: new Abstract: Robots in shared spaces often move in ways that are difficult for people to interpret, placing the burden on humans to adapt. High-DoF robots exhibit motion that people read as expressive, intentionally or not, making it important to understand how such cues are perceived. We present an online video study evaluating how different signaling modalities, expressive motion, lights, text, and audio, shape people's ability to understand a quadruped robot's upcoming navigation actions (Boston Dynamics Spot). Across four common scenarios, we measure how each modality influences humans' (1) accuracy in predicting the robot's next navigation action, (2) confidence in that prediction, and (3) trust in the robot to act safely. The study tests how expressive motions compare to explicit channels, whether ...

• 1.Anticipatory Reinforcement Learning: From Generative Path-Laws to Distributional Value Functions

  This paper introduces Anticipatory Reinforcement Learning (ARL), a novel framework designed to bridge the gap between non-Markovian decision processes and classical reinforcement learning architectures, specifically under the constraint of a single observed trajectory. In environments characterised by jump-diffusions and structural breaks, traditional state-based methods often fail to capture the essential path-dependent geometry required for accurate foresight. We resolve this by lifting the state space into a signature-augmented manifold, where the history of the process is embedded as a dynamical coordinate. By utilising a self-consistent field approach, the agent maintains an anticipated proxy of the future path-law, allowing for a deterministic evaluation of expected returns. This transition from stochastic branching to a single-pass...

• 2.Transfer Learning for Loan Recovery Prediction under Distribution Shifts with Heterogeneous Feature Spaces

  Accurate forecasting of recovery rates (RR) is central to credit risk management and regulatory capital determination. In many loan portfolios, however, RR modeling is constrained by data scarcity arising from infrequent default events. Transfer learning (TL) offers a promising avenue to mitigate this challenge by exploiting information from related but richer source domains, yet its effectiveness critically depends on the presence and strength of distributional shifts, and on potential heterogeneity between source and target feature spaces. This paper introduces FT-MDN-Transformer, a mixture-density tabular Transformer architecture specifically designed for TL in RR forecasting across heterogeneous feature sets. The model produces both loan-level point estimates and portfolio-level predictive distributions, thereby supporting a wide ra...

• 3.Financial Anomaly Detection for the Canadian Market

  In this work we evaluate the performance of three classes of methods for detecting financial anomalies: topological data analysis (TDA), principal component analyis (PCA), and Neural Network-based approaches. We apply these methods to the TSX-60 data to identify major financial stress events in the Canadian stock market. We show how neural network-based methods (such as GlocalKD and One-Shot GIN(E)) and TDA methods achieve the strongest performance. The effectiveness of TDA in detecting financial anomalies suggests that global topological properties are meaningful in distinguishing financial stress events.

• 4.Reinforcement Learning for Speculative Trading under Exploratory Framework

  We study a speculative trading problem within the exploratory reinforcement learning (RL) framework of Wang et al. [2020]. The problem is formulated as a sequential optimal stopping problem over entry and exit times under general utility function and price process. We first consider a relaxed version of the problem in which the stopping times are modeled by the jump times of Cox processes driven by bounded, non-randomized intensity controls. Under the exploratory formulation, the agent's randomized control is characterized via the probability measure over the jump intensities, and their objective function is regularized by Shannon's differential entropy. This yields a system of the exploratory HJB equations and Gibbs distributions in closed-form as the optimal policy. Error estimates and convergence of the RL objective to the value functi...

• 5.JetPrism: diagnosing convergence for generative simulation and inverse problems in nuclear physics

  High-fidelity Monte Carlo simulations and complex inverse problems, such as mapping smeared experimental observations to ground-truth states, are computationally intensive yet essential for robust data analysis. Conditional Flow Matching (CFM) offers a mathematically robust approach to accelerating these tasks, but we demonstrate its standard training loss is fundamentally misleading. In rigorous physics applications, CFM loss plateaus prematurely, serving as an unreliable indicator of true convergence and physical fidelity. To investigate this disconnect, we designed JetPrism, a configurable CFM framework acting as an efficient generative surrogate for evaluating unconditional generation and conditional detector unfolding. Using synthetic stress tests and a Jefferson Lab kinematic dataset ($γp \to ρ^0 p \to π^+π^- p$) relevant to the for...

• 1.From Rich Text to Video: RCS Universal Profile 4.0 has arrived

  Summary available at source link.

• 2.Mobile Money accounted for $2 trillion in  transactions in 2025, doubling since 2021 as active accounts continue to grow

  Summary available at source link.

• 3.Strengthening the Global Fight Against Fraud and Scams – Takeaways from the Global Fraud Summit in Vienna

  Summary available at source link.

• 4.GSMA MWC26 Barcelona closes 20th anniversary edition

  Summary available at source link.

• 5.From Ambition to Execution: How Open Gateway Is Scaling the Global API Economy

  Summary available at source link.

• 1.Rethinking Model Efficiency: Multi-Agent Inference with Large Models

  Most vision-language models (VLMs) apply a large language model (LLM) as the decoder, where the response tokens are generated sequentially through autoregression. Therefore, the number of output tokens can be the bottleneck of the end-to-end latency. However, different models may require vastly different numbers of output tokens to achieve comparable performance. In this work, we conduct a comprehensive analysis of the latency across different components of VLMs on simulated data. The experiment shows that a large model with fewer output tokens can be more efficient than a small model with a long output sequence. The empirical study on diverse real-world benchmarks confirms the observation that a large model can achieve better or comparable performance as a small model with significantly fewer output tokens. To leverage the efficiency of ...

• 2.TriAttention: Efficient Long Reasoning with Trigonometric KV Compression

  Extended reasoning in large language models (LLMs) creates severe KV cache memory bottlenecks. Leading KV cache compression methods estimate KV importance using attention scores from recent post-RoPE queries. However, queries rotate with position during RoPE, making representative queries very few, leading to poor top-key selection and unstable reasoning. To avoid this issue, we turn to the pre-RoPE space, where we observe that Q and K vectors are highly concentrated around fixed non-zero centers and remain stable across positions -- Q/K concentration. We show that this concentration causes queries to preferentially attend to keys at specific distances (e.g., nearest keys), with the centers determining which distances are preferred via a trigonometric series. Based on this, we propose TriAttention to estimate key importance by leveraging ...

• 3.Rethinking Exploration in RLVR: From Entropy Regularization to Refinement via Bidirectional Entropy Modulation

  Reinforcement learning with verifiable rewards (RLVR) has significantly advanced the reasoning capabilities of large language models (LLMs). However, it faces a fundamental limitation termed \textit{restricted exploration}, where the policy rapidly converges to a narrow set of solutions. While entropy regularization is a popular approach used to sustain exploration, it often proves unreliable for LLMs, suffering from high hyperparameter sensitivity and yielding only marginal performance gains. Motivated by these inefficiencies, we propose to rethink the relationship between policy entropy and exploration. By deriving a parametric formulation of group-relative advantage estimation and analyzing entropy dynamics, we conceptually decompose policy entropy into \textit{informative entropy}, which preserves diverse solution paths, and \textit{s...

• 4.Optimizing LLM Prompt Engineering with DSPy Based Declarative Learning

  Large Language Models (LLMs) have shown strong performance across a wide range of natural language processing tasks; however, their effectiveness is highly dependent on prompt design, structure, and embedded reasoning signals. Conventional prompt engineering methods largely rely on heuristic trial-and-error processes, which limits scalability, reproducibility, and generalization across tasks. DSPy, a declarative framework for optimizing text-processing pipelines, offers an alternative approach by enabling automated, modular, and learnable prompt construction for LLM-based systems.This paper presents a systematic study of DSPy-based declarative learning for prompt optimization, with emphasis on prompt synthesis, correction, calibration, and adaptive reasoning control. We introduce a unified DSPy LLM architecture that combines symbolic plan...

• 5.Assessing Large Language Models for Stabilizing Numerical Expression in Scientific Software

  Scientific software relies on high-precision computation, yet finite floating-point representations can introduce precision errors that propagate in safety-critical domains. Despite the growing use of large language models (LLMs) in scientific applications, their reliability in handling floating-point numerical stability has not been systematically evaluated. This paper evaluates LLMs' reasoning on high-precision numerical computation through two numerical stabilization tasks: (1) detecting instability in numerical expressions by generating error-inducing inputs (detection), and (2) rewriting expressions to improve numerical stability (stabilization). Using popular numerical benchmarks, we assess six LLMs on nearly 2,470 numerical structures, including nested conditionals, high-precision literals, and multi-variable arithmetic. Our result...

• 1.Vanast: Virtual Try-On with Human Image Animation via Synthetic Triplet Supervision

  We present Vanast, a unified framework that generates garment-transferred human animation videos directly from a single human image, garment images, and a pose guidance video. Conventional two-stage pipelines treat image-based virtual try-on and pose-driven animation as separate processes, which often results in identity drift, garment distortion, and front-back inconsistency. Our model addresses these issues by performing the entire process in a single unified step to achieve coherent synthesis. To enable this setting, we construct large-scale triplet supervision. Our data generation pipeline includes generating identity-preserving human images in alternative outfits that differ from garment catalog images, capturing full upper and lower garment triplets to overcome the single-garment-posed video pair limitation, and assembling diverse i...

• 2.Rethinking Model Efficiency: Multi-Agent Inference with Large Models

  Most vision-language models (VLMs) apply a large language model (LLM) as the decoder, where the response tokens are generated sequentially through autoregression. Therefore, the number of output tokens can be the bottleneck of the end-to-end latency. However, different models may require vastly different numbers of output tokens to achieve comparable performance. In this work, we conduct a comprehensive analysis of the latency across different components of VLMs on simulated data. The experiment shows that a large model with fewer output tokens can be more efficient than a small model with a long output sequence. The empirical study on diverse real-world benchmarks confirms the observation that a large model can achieve better or comparable performance as a small model with significantly fewer output tokens. To leverage the efficiency of ...

• 3.HI-MoE: Hierarchical Instance-Conditioned Mixture-of-Experts for Object Detection

  Mixture-of-Experts (MoE) architectures enable conditional computation by activating only a subset of model parameters for each input. Although sparse routing has been highly effective in language models and has also shown promise in vision, most vision MoE methods operate at the image or patch level. This granularity is poorly aligned with object detection, where the fundamental unit of reasoning is an object query corresponding to a candidate instance. We propose Hierarchical Instance-Conditioned Mixture-of-Experts (HI-MoE), a DETR-style detection architecture that performs routing in two stages: a lightweight scene router first selects a scene-consistent expert subset, and an instance router then assigns each object query to a small number of experts within that subset. This design aims to preserve sparse computation while better matchi...

• 4.HorizonWeaver: Generalizable Multi-Level Semantic Editing for Driving Scenes

  Ensuring safety in autonomous driving requires scalable generation of realistic, controllable driving scenes beyond what real-world testing provides. Yet existing instruction guided image editors, trained on object-centric or artistic data, struggle with dense, safety-critical driving layouts. We propose HorizonWeaver, which tackles three fundamental challenges in driving scene editing: (1) multi-level granularity, requiring coherent object- and scene-level edits in dense environments; (2) rich high-level semantics, preserving diverse objects while following detailed instructions; and (3) ubiquitous domain shifts, handling changes in climate, layout, and traffic across unseen environments. The core of HorizonWeaver is a set of complementary contributions across data, model, and training: (1) Data: Large-scale dataset generation, where we ...

• 5.Learning, Potential, and Retention: An Approach for Evaluating Adaptive AI-Enabled Medical Devices

  This work addresses challenges in evaluating adaptive artificial intelligence (AI) models for medical devices, where iterative updates to both models and evaluation datasets complicate performance assessment. We introduce a novel approach with three complementary measurements: learning (model improvement on current data), potential (dataset-driven performance shifts), and retention (knowledge preservation across modification steps), to disentangle performance changes caused by model adaptations versus dynamic environments. Case studies using simulated population shifts demonstrate the approach's utility: gradual transitions enable stable learning and retention, while rapid shifts reveal trade-offs between plasticity and stability. These measurements provide practical insights for regulatory science, enabling rigorous assessment of the saf...

• 1.Why AI Systems Fail Quietly

  In late-stage testing of a distributed AI platform, engineers sometimes encounter a perplexing situation: every monitoring dashboard reads “healthy,” yet users report that the system’s decisions are slowly becoming wrong. Engineers are trained to recognize failure in familiar ways: a service crashes, a sensor stops responding, a constraint violation triggers a shutdown. Something breaks, and the system tells you. But a growing class of software failures looks very different. The system keeps running, logs appear normal, and monitoring dashboards stay green. Yet the system’s behavior quietly drifts away from what it was designed to do. This pattern is becoming more common as autonomy spreads across software systems. Quiet failure is emerging as one of the defining engineering challenges of autonomous systems because correctness now depends...

• 2.AI Is Insatiable

  While browsing our website a few weeks ago, I stumbled upon “ How and When the Memory Chip Shortage Will End ” by Senior Editor Samuel K. Moore. His analysis focuses on the current DRAM shortage caused by AI hyperscalers’ ravenous appetite for memory, a major constraint on the speed at which large language models run. Moore provides a clear explanation of the shortage, particularly for high bandwidth memory (HBM). As we and the rest of the tech media have documented, AI is a resource hog. AI electricity consumption could account for up to 12 percent of all U.S. power by 2028. Generative AI queries consumed 15 terawatt-hours in 2025 and are projected to consume 347 TWh by 2030. Water consumption for cooling AI data centers is predicted to double or even quadruple by 2028 compared to 2023. But Moore’s reporting shines a light on an obscure ...

• 3.The AI Data Centers That Fit on a Truck

  A traditional data center protects the expensive hardware inside it with a “shell” constructed from steel and concrete. Constructing a data center’s shell is inexpensive compared to the cost of the hardware and infrastructure inside it, but it’s not trivial. It takes time for engineers to consider potential sites, apply for permits, and coordinate with construction contractors. That’s a problem for those looking to quickly deploy AI hardware, which has led companies like Duos Edge AI and LG CNS to respond with a more modular approach. They use pre-fabricated, self-contained boxes that can be deployed in months instead of years. The boxes can operate alone or in tandem with others, providing the option to add more if required. “I just came back from Nvidia’s GTC, and a lot of [companies] are sitting on their deployment because their data c...

• 4.Why Are Large Language Models so Terrible at Video Games?

  Large language models (LLMs) have improved so quickly that the benchmarks themselves have evolved, adding more complex problems in an effort to challenge the latest models. Yet LLMs haven’t improved across all domains, and one task remains far outside their grasp: They have no idea how to play video games. While a few have managed to beat a few games (for example, Gemini 2.5 Pro beat Pokemon Blue in May of 2025), these exceptions prove the rule. The eventually victorious AI completed games far more slowly than a typical human player, made bizarre and often repetitive mistakes, and required custom software to guide their interactions with the game. Julian Togelius , the director of New York University’s Game Innovation Lab and co-founder of AI game testing company Modl.ai, explored the implications of LLMs’ limitations in video games in a ...

• 5.How NYU’s Quantum Institute Bridges Science and Application

  This sponsored article is brought to you by NYU Tandon School of Engineering . Within a 6 mile radius of New York University’s (NYU) campus, there are more than 500 tech industry giants, banks, and hospitals. This isn’t just a fact about real estate, it’s the foundation for advancing quantum discovery and application. While the world races to harness quantum technology, NYU is betting that the ultimate advantage lies not solely in a lab, but in the dense, demanding, and hyper-connected urban ecosystem that surrounds it. With the launch of its NYU Quantum Institute (NYUQI), NYU is positioning itself as the central node in this network; a “full stack” powerhouse built on the conviction that it has found the right place, and the right time, to turn quantum science into tangible reality. Proximity advantage is essential because quantum scienc...

• 1.Disintegrating the Org Chart: ServiceNow’s Jacqui Canney

  In this episode of the Me, Myself, and AI podcast, Sam Ransbotham is joined by Jacqui Canney, chief people and AI enablement officer at ServiceNow. Jacqui outlines how the software company has embedded AI agents into processes like employee onboarding to automate tasks, personalize experiences, and free up people’s time to focus on higher-value work. […]

• 2.How to Reap Compound Benefits From Generative AI

  Carolyn Geason-Beissel/MIT SMR | Minneapolis Institute of Art In domain after domain, AI has compressed work that used to be expensive — generating drafts, code, prototypes, and analyses. The marginal cost of a first attempt has dropped sharply. What remains expensive is what happens after the output arrives: evaluating what gets generated. That involves separating […]

• 3.Job Pivots in the Age of AI: Lessons From Mike Mulligan and His Steam Shovel

  Matt Harrison Clough As organizations like Amazon, PwC, and Microsoft have announced AI-fueled layoffs, it’s no surprise that half of Americans have expressed concern about AI’s larger potential impact on their jobs. Of course, companies can attribute layoffs to AI efficiencies while trimming workforces for various reasons. Yet there is no question that artificial intelligence […]

• 4.The Best Customers to Study When Scaling Into a New Market

  Carolyn Geason-Beissel/MIT SMR | Getty Images For tech companies worldwide, expanding into a new market is both a rite of passage and a moment of truth. It represents the transition from early promise to meaningful scale — an opportunity to increase revenue, signal growth potential to investors, and unlock powerful sources of differentiation, such as […]

• 5.Level Up Your Crisis Management Skills

  Michael Austin/theispot.com The Research The authors conducted in-depth interviews with senior leaders with direct experience guiding large, complex systems through unexpected shocks. Their sample included a former prime minister, CEOs, board chairs and directors of multinational corporations, a central bank governor, a national chief of defense, and a national fire marshal. Participants represented a diversity […]

• 1.Can Personal Access to Medical Expertise Overcome Vaccine Hesitancy? -- by D. Mark Anderson, Ron Diris, Raymond Montizaan, Daniel I. Rees

  Using data on applicants to Dutch medical schools and their older relatives (i.e., parents, aunts, and uncles ages 60+), we estimate the effect of personal access to medical expertise on vaccine hesitancy. Leveraging variation in lottery outcomes that determine admission to medical schools, we find that having a physician in the family increases the likelihood of complying with government recommendations that anyone over the age of 59 receive a second booster dose of a COVID-19 vaccine. Our estimated effects are strongest for having a female physician in the family, suggesting important gender-based differences in how medical expertise is communicated.

• 2.Why Do Americans No Longer Work So Much More Than Non-Americans? -- by Serdar Birinci, Loukas Karabarbounis, Kurt See

  In the 1990s, Americans used to work much more than non-Americans. Nowadays, about half of the gap in hours worked has reversed. To evaluate the convergence of working hours, we develop a tractable model of labor supply enriched with multiple sources of heterogeneity across individuals, an extensive margin of participation, multi-member households, and an elaborate system of taxes and benefits upon non-employment. Using detailed measurements from micro-level and aggregate datasets, we identify model parameters and sources of heterogeneity across individuals for various countries. We run a horse race between competing explanations and find that U.S. hours per person declined after 2000 owing mainly to the rise of government health benefits provided to the non-employed. Non-U.S. countries have generous benefits for the non-employed, but th...

• 3.AI Patents in the United States and China: Measurement, Organization, and Knowledge Flows -- by Hanming Fang, Xian Gu, Hanyin Yan, Wu Zhu

  We develop a high-precision classifier to measure artificial intelligence (AI) patents by fine-tuning PatentSBERTa on manually labeled data from the USPTO’s AI Patent Dataset. Our classifier substantially improves the existing USPTO approach, achieving 97.0% precision, 91.3% recall, and a 94.0% F1 score, and it generalizes well to Chinese patents based on citation and lexical validation. Applying it to granted U.S. patents (1976–2023) and Chinese patents (2010–2023), we document rapid growth in AI patenting in both countries and broad convergence in AI patenting intensity and subfield composition, even as China surpasses the United States in recent annual patent counts. The organization of AI innovation nevertheless differs sharply: U.S. AI patenting is concentrated among large private incumbents and established hubs, whereas Chinese AI p...

• 4.Tariffs, Global Value Chains, and the Incidence of Protection: Evidence from US Automobiles -- by Luke Heeney, Christopher R. Knittel, Jasdeep Mandia

  In many modern industries, firms compete in differentiated-product markets while relying on complex global value chains for intermediate inputs. In such settings, trade policies such as tariffs on vehicles and parts operate not only through consumer substitution and firm pricing, but also through firms’ cost structures and sourcing decisions. We develop a structural model of the U.S. automobile market that integrates random-coefficients demand, multiproduct firm pricing, and a flexible supply-side framework in which shocks to the cost of imported parts transmit imperfectly into manufacturers’ marginal costs. The model is disciplined by novel model-level data on imported-parts exposure and exploits exchange-rate variation to identify cost pass-through. Our counterfactual analysis quantifies the effects of alternative tariff policies on pri...

• 5.Learning How To Borrow in a Fintech World: Consumer Behavior When Search Costs Are (Near) Zero -- by Alex Günsberg, Camelia M. Kuhnen

  Online loan marketplaces are changing consumer lending. Here we investigate consumer behavior in these markets with near-zero search costs. Using administrative data on 730,000 applications, 750,000 offers, and 200,000 individuals, together with credit registry records, we document four facts. First, substantial within-applicant dispersion in offered terms makes search highly valuable. Second, marketplace nudges mitigate choice complexity. Third, applicants search significantly, applying repeatedly, asking for different terms, and rejecting offers, in ways consistent with their creditworthiness. Fourth, dynamic adverse selection constrains search, as lenders penalize repeat applicants. Our findings highlight trade-offs between informational gains from search, and reputational and cognitive costs.

• 1.The Fed Has Two Tools to Influence Money Market Conditions 

  The Federal Reserve’s 2022-23 tightening cycle involved the use of two monetary policy tools: changes in administrative rates and changes in the size of its balance sheet. This post highlights the results of a recent Staff Report that explores how these tools affect money market conditions. Using confidential trade-level data, we find that both tools have significant effects on the pricing of funds sourced through repo. These results suggest that the Fed can manage how financing conditions are affected even as it influences economic conditions. For example, the Fed can lower its administrative rates to loosen economic conditions, while shrinking its balance sheet to maintain financing conditions in the money markets. 

• 2.Treasury Market Liquidity Since April 2025

  In this post, we examine the evolution of U.S. Treasury market liquidity over the past year, which has witnessed myriad economic and political developments. Liquidity worsened markedly one year ago as volatility increased following the announcement of higher-than-expected tariffs. Liquidity quickly improved when the tariff increases were partially rolled back and then remained fairly stable thereafter (through the end of our sample in February 2026), including after the recent Supreme Court decision striking down the emergency tariffs and the subsequent announcement of new tariffs.

• 3.Behind the ATM: Exploring the Structure of Bank Holding Companies

  Many modern banking organizations are highly complex. A “bank” is often a larger structure made up of distinct entities, each subject to different regulatory, supervisory, and reporting requirements. For researchers and policymakers, understanding how these institutions are structured and how they have evolved over time is essential. In this post, we illustrate what a modern financial holding company looks like in practice, document how banks’ organizational structures have changed over time, and explain why these details matter for conducting accurate analyses of the financial system.

• 4.Sports Betting Is Everywhere, Especially on Credit Reports

  Since 2018, more than thirty states have legalized mobile sports betting, leading to more than a half trillion dollars in wagers. In our recent Staff Report, we examine how legalized sports betting affects household financial health by comparing betting activity and consumer credit outcomes between states that legalized to those that have not. We find that legalization increases spending at online sportsbooks roughly tenfold, but betting does not stop at state boundaries. Nearby areas where betting is not legal still experience roughly 15 percent the increase of counties where it is legal. At the same time, consumer financial health suffers. Our analysis finds rising delinquencies in participating states,...

• 5.China’s Electric Trade

  China has spent considerable government resources to develop advanced electric technology industries, such as those that produce electric vehicles, lithium batteries, and solar panels. These efforts have spilled over to international trade as improvements in price and quality have increased the global demand for these goods. One consequence is that passenger cars and batteries have been disproportionately large contributors to the rise in the country’s trade surplus in recent years. This has not been the case, though, for solar panels, as falling prices due to a supply glut pulled down export revenues despite higher volumes.

• 1.How Women Succeed in Male-Dominated Fields

  Discussions about gender and work often morph into cultural clashes, driven by fear, ideology, or frustration. But recent empirical research suggests that having female peers and professors can help women succeed in the workplace without hurting men.

• 2.The Real AI Race

  While the invention of general-purpose technologies creates new opportunities, it is their diffusion across industries and economies that drives economic transformation. Early evidence suggests that AI is likely to follow a similar trajectory.

• 3.Is Economic Forecasting Still Possible?

  Much of the forecasting profession—and the economic theory that underpins it—still defaults to single baselines that treat the future as a probabilistic replica of the past. But when the structure of the economy changes in unforeseeable ways, as it is now, forecasters must acknowledge that many futures are possible.

• 4.Can India Become the World’s Innovation Capital?

  As supply chains fragment, AI rewrites the rules of competition, and geopolitical volatility intensifies, companies are searching for stable, scalable, innovation‑rich environments. India can provide what they need, and with the right enabling policies, it can transform its advantages into a lasting global lead.

• 5.A Bad Deal Today Means a Bigger War Tomorrow

  Iran has given the international community little reason to trust it since the 1979 revolution that led to the establishment of the Islamic Republic. Negotiations to end the war today would be dangerously premature, naive, and likely to produce an outcome that all but guarantees another war.

• 1.SpaceX targets record IPO, riding Starlink’s explosive growth

  SpaceX has confidentially filed for what could be the largest IPO in history, with Starlink’s booming satellite business making the case for a cumulative $2 trillion valuation In sum — what to know” SpaceX’s IPO moment: According to reports, SpaceX…

• 2.ZTE targets 5G monetization shift with AI-driven service platform

  ZTE said the solution integrates directly with the 5G core, enabling end-to-end automation and reducing reliance on traditional operational support systems In sum – what to know: Intent-based services – ZTE’s platform enables users to request network performance outcomes directly,…

• 3.The building is the network (part 1) – dumb Wi-Fi is dead (Analyst Angle)

  Multi-dwelling unit (MDU) service providers have spent years building excellent Wi-Fi networks and collecting predictable fees. The ceiling on what that network can generate is gone. For years, the managed Wi-Fi industry sold itself on a straightforward value proposition: take…

• 4.Global telecom capex set to fall in 2026: Dell’Oro

  Dell’Oro forecasts a 2% decline in global telecom capex in 2026, followed by modest growth at a CAGR of around 1% through 2030 In sum – what to know: Capex set to decline – Global telecom capex is expected to…

• 5.ABI on AI infra | AI grid may be the next telecoms arms race (Analyst Angle)

  Telcos are exploring NVIDIA’s AI grid, but edge GPU deployment lacks a strong latency or cost case today; only physical AI use cases justify it, with gradual rollout expected toward future 6G networks. The diffusion of AI in telco networks…

• 1.Source Error

  Check Feed

• 1.DRL-Based Phase Optimization for O-RIS in Dual-Hop Hard-Switching FSO/RIS-aided RF and UWOC Systems

  This paper presents a dual-hop hybrid framework that integrates a free-space optical (FSO)/RIS-aided radio frequency (RF) link operating under a hard-switching protocol as the first hop, and an optical reconfigurable intelligent surface (O-RIS)-assisted underwater wireless optical communication (UWOC) link as the second hop. To capture realistic underwater dynamics, the Oceanic Turbulence Optical Power Spectrum (OTOPS) is employed for accurate turbulence modeling. For efficient O-RIS phase control, deep reinforcement learning (DRL) algorithms, specifically the Deep Deterministic Policy Gradient (DDPG) and Twin Delayed DDPG (TD3), have been developed to optimize the phase shifts of O-RIS elements. Simulation results demonstrate that the proposed system substantially improves outage probability and channel capacity, with TD3 achieving super...

• 2.Pinching Antenna Systems (PASS): Enabling Reconfigurable and Controllable Wireless Channels -- A Comprehensive Survey

  The evolution of wireless networks is driving new paradigms for consideration in upcoming generations. To this end, the 6G anticipates the development of several data-rate-hungry applications, in addition to a forecast growth in sensing-centric applications. Such an evolution, however, is unbalanced on the other side by the accentuated scarcity of spectrum, which opens up urgent needs to develop spectrum-efficient communication and sensing techniques. Due to the inability of the traditional multi-antenna schemes to enhance a wireless channel quality, increasing interest has been paid to wireless channel-altering schemes, such as reconfigurable intelligent surfaces and movable antennas. Recently, a new technique in this category, called pinching antennas (PAs), was introduced and tested. PA systems (PASS) are based on extending the reach o...

• 3.Merkle Tree Certificate Post-Quantum PKI for Kubernetes and Cloud-Native 5G/B5G Core

  Post-quantum signature schemes such as ML-DSA-65 produce signatures of 3,309 bytes and public keys of 1,952 bytes over 50 times larger than classical Ed25519. In TLS-authenticated environments like Kubernetes control planes and 5G Core networks, where every inter-component connection is mutually authenticated, this overhead compounds across thousands of handshakes per second. Merkle Tree Certificates (MTC), currently under development at IETF, replace per-certificate issuer signatures with Merkle inclusion proofs and, in the landmark mode, eliminate on-wire signatures from certificate authentication entirely. We present MTC-based PKI architectures for Kubernetes and 3GPP 5G Service-Based Architecture. Starting from the infrastructure layer, we replace the Kubernetes cluster CA with an MTCA deployment that issues MTC certificates to contro...

• 4.Advanced Holographic Multi-Antenna Solutions for Global Non-Terrestrial Network Integration in IMT-2030 Systems

  Sixth-generation (6G) networks are expected to provide ubiquitous connectivity across terrestrial and non-terrestrial domains. This will be possible by integrating non-terrestrial networks (NTNs) to extend coverage to underserved areas. Antennas are central to this vision, with multiple-input multiple-output (MIMO) technologies receiving the most attention due to their ability to exploit spatial multiplexing to improve link capacity and reliability. However, conventional MIMO can consume significant energy, as each antenna element typically requires an independent RF chain. This limitation is particularly critical in non-terrestrial systems, where onboard energy resources are limited. Holographic MIMO (HMIMO) has emerged as a promising alternative in this context. These systems are based on theoretically continuous apertures, where radiat...

• 5.Enhancing 6G Wireless Intelligence: Do LLMs Work for CSI Prediction?

  In high-mobility 6G scenarios, rapidly time-varying channels lead to very short coherence times, which makes conventional pilot-based channel state information (CSI) estimation approaches prone to outdated information or excessive pilot overhead. Therefore, channel prediction becomes essential in such dynamic wireless systems. To address this challenge, large language models (LLMs) are emerging learning frameworks that have recently attracted attention for CSI prediction due to their strong sequence modeling capability and ability to generalize across different environments. This paper proposes an LLM-based framework for channel prediction in high-mobility orthogonal time frequency space (OTFS) communication systems. In this work, we develop a physics-aware LLM-based predictor that learns the temporal evolution of OTFS channel coefficient...

• 1.Anticipatory Reinforcement Learning: From Generative Path-Laws to Distributional Value Functions

  This paper introduces Anticipatory Reinforcement Learning (ARL), a novel framework designed to bridge the gap between non-Markovian decision processes and classical reinforcement learning architectures, specifically under the constraint of a single observed trajectory. In environments characterised by jump-diffusions and structural breaks, traditional state-based methods often fail to capture the essential path-dependent geometry required for accurate foresight. We resolve this by lifting the state space into a signature-augmented manifold, where the history of the process is embedded as a dynamical coordinate. By utilising a self-consistent field approach, the agent maintains an anticipated proxy of the future path-law, allowing for a deterministic evaluation of expected returns. This transition from stochastic branching to a single-pass...

• 2.SoK: Blockchain Agent-to-Agent Payments

  Agentic AI rivals human capabilities across a wide range of domains. Looking ahead, it is foreseeable that AI agents will autonomously handle complex workflows and interactions. Early prototypes of this paradigm are emerging, e.g., OpenClaw and Moltbook, signaling a shift toward Agent-to-Agent (A2A) ecosystems. However, despite these promising blueprints, critical trust and security challenges remain, particularly in scenarios involving financial transactions. Ensuring secure and reliable payment mechanisms between unknown and untrusted agents is crucial to complete a fully functional and trustworthy A2A ecosystem. Although blockchain-based infrastructures provide a natural foundation for this setting, via programmable settlement, transparent accounting, and open interoperability, trust and security challenges have not yet been fully addr...

• 3.Adaptive VaR Control for Standardized Option Books under Marking Frictions

  Short-horizon risk control matters for hedging and capital allocation. Yet existing Value-at-Risk studies rarely address standardized option books or the next-day valuation frictions that arise in derivatives data. This paper develops a framework for tail-risk control in standardized option books. The analysis focuses on the next-day realized loss and combines a base conditional quantile forecast with sequential conformal recalibration for adaptive Value-at-Risk control. This design addresses two central difficulties: unstable tail-risk forecasts under changing market conditions and the practical challenge of next-day valuation when exact same-contract quotes are unavailable. It also preserves economic interpretability through standardized construction and spot hedging when needed. Using SPX option data from 2018 to 2025, we show that t...

• 4.Debiasing LLMs by Fine-tuning

  Prior research shows that large language models (LLMs) exhibit systematic extrapolation bias when forming predictions from both experimental and real-world data, and that prompt-based approaches appear limited in alleviating this bias. We propose a supervised fine-tuning (SFT) approach that uses Low-Rank Adaptation (LoRA) to train off-the-shelf LLMs on instruction datasets constructed from rational benchmark forecasts. By intervening at the parameter level, SFT changes how LLMs map observed information into forecasts and thereby mitigates extrapolation bias. We evaluate the fine-tuned model in two settings: controlled forecasting experiments and cross-sectional stock return prediction. In both settings, fine-tuning corrects the extrapolative bias out-of-sample, establishing a low-cost and generalizable method for debiasing LLMs.

• 5.Financial Anomaly Detection for the Canadian Market

  In this work we evaluate the performance of three classes of methods for detecting financial anomalies: topological data analysis (TDA), principal component analyis (PCA), and Neural Network-based approaches. We apply these methods to the TSX-60 data to identify major financial stress events in the Canadian stock market. We show how neural network-based methods (such as GlocalKD and One-Shot GIN(E)) and TDA methods achieve the strongest performance. The effectiveness of TDA in detecting financial anomalies suggests that global topological properties are meaningful in distinguishing financial stress events.

• 1.DRL-Based Phase Optimization for O-RIS in Dual-Hop Hard-Switching FSO/RIS-aided RF and UWOC Systems

  This paper presents a dual-hop hybrid framework that integrates a free-space optical (FSO)/RIS-aided radio frequency (RF) link operating under a hard-switching protocol as the first hop, and an optical reconfigurable intelligent surface (O-RIS)-assisted underwater wireless optical communication (UWOC) link as the second hop. To capture realistic underwater dynamics, the Oceanic Turbulence Optical Power Spectrum (OTOPS) is employed for accurate turbulence modeling. For efficient O-RIS phase control, deep reinforcement learning (DRL) algorithms, specifically the Deep Deterministic Policy Gradient (DDPG) and Twin Delayed DDPG (TD3), have been developed to optimize the phase shifts of O-RIS elements. Simulation results demonstrate that the proposed system substantially improves outage probability and channel capacity, with TD3 achieving super...

• 2.Pinching Antenna Systems (PASS): Enabling Reconfigurable and Controllable Wireless Channels -- A Comprehensive Survey

  The evolution of wireless networks is driving new paradigms for consideration in upcoming generations. To this end, the 6G anticipates the development of several data-rate-hungry applications, in addition to a forecast growth in sensing-centric applications. Such an evolution, however, is unbalanced on the other side by the accentuated scarcity of spectrum, which opens up urgent needs to develop spectrum-efficient communication and sensing techniques. Due to the inability of the traditional multi-antenna schemes to enhance a wireless channel quality, increasing interest has been paid to wireless channel-altering schemes, such as reconfigurable intelligent surfaces and movable antennas. Recently, a new technique in this category, called pinching antennas (PAs), was introduced and tested. PA systems (PASS) are based on extending the reach o...

• 3.Merkle Tree Certificate Post-Quantum PKI for Kubernetes and Cloud-Native 5G/B5G Core

  Post-quantum signature schemes such as ML-DSA-65 produce signatures of 3,309 bytes and public keys of 1,952 bytes over 50 times larger than classical Ed25519. In TLS-authenticated environments like Kubernetes control planes and 5G Core networks, where every inter-component connection is mutually authenticated, this overhead compounds across thousands of handshakes per second. Merkle Tree Certificates (MTC), currently under development at IETF, replace per-certificate issuer signatures with Merkle inclusion proofs and, in the landmark mode, eliminate on-wire signatures from certificate authentication entirely. We present MTC-based PKI architectures for Kubernetes and 3GPP 5G Service-Based Architecture. Starting from the infrastructure layer, we replace the Kubernetes cluster CA with an MTCA deployment that issues MTC certificates to contro...

• 4.Advanced Holographic Multi-Antenna Solutions for Global Non-Terrestrial Network Integration in IMT-2030 Systems

  Sixth-generation (6G) networks are expected to provide ubiquitous connectivity across terrestrial and non-terrestrial domains. This will be possible by integrating non-terrestrial networks (NTNs) to extend coverage to underserved areas. Antennas are central to this vision, with multiple-input multiple-output (MIMO) technologies receiving the most attention due to their ability to exploit spatial multiplexing to improve link capacity and reliability. However, conventional MIMO can consume significant energy, as each antenna element typically requires an independent RF chain. This limitation is particularly critical in non-terrestrial systems, where onboard energy resources are limited. Holographic MIMO (HMIMO) has emerged as a promising alternative in this context. These systems are based on theoretically continuous apertures, where radiat...

• 5.Enhancing 6G Wireless Intelligence: Do LLMs Work for CSI Prediction?

  In high-mobility 6G scenarios, rapidly time-varying channels lead to very short coherence times, which makes conventional pilot-based channel state information (CSI) estimation approaches prone to outdated information or excessive pilot overhead. Therefore, channel prediction becomes essential in such dynamic wireless systems. To address this challenge, large language models (LLMs) are emerging learning frameworks that have recently attracted attention for CSI prediction due to their strong sequence modeling capability and ability to generalize across different environments. This paper proposes an LLM-based framework for channel prediction in high-mobility orthogonal time frequency space (OTFS) communication systems. In this work, we develop a physics-aware LLM-based predictor that learns the temporal evolution of OTFS channel coefficient...

• 1.Advanced Holographic Multi-Antenna Solutions for Global Non-Terrestrial Network Integration in IMT-2030 Systems

  Sixth-generation (6G) networks are expected to provide ubiquitous connectivity across terrestrial and non-terrestrial domains. This will be possible by integrating non-terrestrial networks (NTNs) to extend coverage to underserved areas. Antennas are central to this vision, with multiple-input multiple-output (MIMO) technologies receiving the most attention due to their ability to exploit spatial multiplexing to improve link capacity and reliability. However, conventional MIMO can consume significant energy, as each antenna element typically requires an independent RF chain. This limitation is particularly critical in non-terrestrial systems, where onboard energy resources are limited. Holographic MIMO (HMIMO) has emerged as a promising alternative in this context. These systems are based on theoretically continuous apertures, where radiat...

• 2.Reimagining RAN Automation in 6G: An Agentic AI Framework with Hierarchical Online Decision Transformer

  In this paper, we propose an Agentic Artificial Intelligence (AI) framework for wireless networks. The framework coordinates a pool of AI agents guided by Natural Language (NL) inputs from a human operator. At its core, the super agent is powered by a Hierarchical Online Decision Transformer (H-ODT). It orchestrates three categories of agents: (i) inter-slice, intra-slice resource allocation agents, (ii) network application orchestration agents, and (iii) self-healing agents. The orchestration takes place with the help of an Agentic Retrieval-Augmented Generation (RAG) module that integrates knowledge from heterogeneous sources. In this proposed methodology, the super agent directly interfaces with operators and generates sequential policies to activate relevant agents. The proposed framework is evaluated against three state-of-the-art ba...

• 3.RL-Loop: Reinforcement Learning-Driven Real-Time 5G Slice Control for Connected and Autonomous Mobility Services

  Smart and connected mobility systems rely on 5G edge infrastructure to support real-time communication, control, and service differentiation. Achieving this requires adaptive resource management mechanisms that can react to rapidly changing traffic conditions. In this paper, we propose RL-Loop, a closed-loop reinforcement learning framework for real-time CPU resource control in 5G network slicing environments supporting connected mobility services. RL-Loop employs a Proximal Policy Optimization (PPO) agent that continuously observes slice-level key performance indicators and adjusts edge CPU allocations at one-second granularity on a real testbed. The framework leverages real-time observability and feedback to enable adaptive, software-defined edge intelligence. Experimental results suggest that RL-Loop can reduce average CPU allocation b...

• 4.CIVIC: Cooperative Immersion Via Intelligent Credit-sharing in DRL-Powered Metaverse

  The Metaverse faces complex resource allocation challenges due to diverse Virtual Environments (VEs), Digital Twins (DTs), dynamic user demands, and strict immersion needs. This paper introduces CIVIC (Cooperative Immersion Via Intelligent Credit-sharing), a novel framework optimizing resource sharing among multiple Metaverse Service Providers (MSPs) to enhance user immersion. Unlike existing methods, CIVIC integrates VE rendering, DT synchronization, credit sharing, and immersion-aware provisioning within a cooperative multi-MSP model. The resource allocation problem is formulated as two NP-hard challenges: a non-cooperative setting where MSPs operate independently and a cooperative setting utilizing a General Credit Pool (GCP) for dynamic resource sharing. Using Deep Reinforcement Learning (DRL) for tuning resources and managing coopera...

• 5.Adversarial Attacks in AI-Driven RAN Slicing: SLA Violations and Recovery

  Next-generation (NextG) cellular networks are designed to support emerging applications with diverse data rate and latency requirements, such as immersive multimedia services and large-scale Internet of Things deployments. A key enabling mechanism is radio access network (RAN) slicing, which dynamically partitions radio resources into virtual resource blocks to efficiently serve heterogeneous traffic classes, including enhanced mobile broadband (eMBB), massive machine-type communications (mMTC), and ultra-reliable low-latency communications (URLLC). In this paper, we study the impact of adversarial attacks on AI-driven RAN slicing decisions, where a budget-constrained adversary selectively jams slice transmissions to bias deep reinforcement learning (DRL)-based resource allocation, and quantify the resulting service level agreement (SLA) ...