ZICQ AI News

ZICQ curates daily AI news and deep dives on models, open-source infrastructure, and developer field reports.

Current filter: Tag: #Adversarial Machine Learning Clear filters

Research & Papers· 2026-11-13#BEST-RQ #Speech Representation Learning #Distributionally Robust Optimization #Discrete Quantization #Self-Supervised Learning

By Mr.Xu0 views0 discussions

Distributionally Robust BEST-RQ Improves Speech Representations Across Codebooks

BEST-RQ learns efficient speech representations from large-scale unlabeled audio through r…

Research & Papers· 2026-11-13

Mr.Xu0 views0 discussions

Frozen Multimodal LLMs Can Power Word-Level Full-Duplex Spoken Dialogue

This paper adapts a frozen general-purpose multimodal LLM into a word-level full-duplex sp…

LLMs & Foundation Models· 2026-11-13

Mr.Xu0 views0 discussions

MDU Benchmark Exposes Multilingual Document-Understanding Gaps That Scale Does Not Close

MDU is a multilingual document-understanding benchmark covering 13 tasks, five languages, …

Agentic· 2026-11-13

Mr.Xu0 views0 discussions

Simulate What-If Users Before Shipping With Persona-Based Agent Experiments

This work presents a framework for persona-based user simulation that formalizes target co…

LLMs & Foundation Models· 2026-11-13

Mr.Xu0 views0 discussions

MITRA-Swarm Decomposes Classical Sanskrit Translation with Explicit Linguistic Structure

MITRA-Swarm decomposes classical Sanskrit machine translation into explicit Sandhi segment…

Research & Papers· 2026-11-13

Mr.Xu0 views0 discussions

Abstention Is Not Compliance: Open Models Ignore Confidence Thresholds

Kalai et al. evaluated four fully open language-model recipes, 12 post-training checkpoint…

Research & Papers· 2026-11-13

Mr.Xu0 views0 discussions

LLM Leaderboards Are Vulnerable to Strategic Voting, Prompting, and Nomination

A study of open LLM arena leaderboards shows that rankings built from anonymous pairwise v…

Agentic· 2026-11-13

Mr.Xu0 views0 discussions

Time Is the Signal: Humans Outperform AI at Adaptive Social Coordination

Researchers evaluated 150 human participants, computer-use agents, and text-prompted langu…

Agentic· 2026-11-13

Mr.Xu0 views0 discussions

SynthCorp Generates Enterprise Data-Agent Benchmarks Through Traceable Event Simulation

SynthCorp creates customizable data estates and benchmarks for LLM-driven data agents by s…

Research & Papers· 2026-11-13

Mr.Xu0 views0 discussions

When Should LLM Semantic Priors Enter Partially-Identified Causal Discovery?

This study separates the use of LLM semantic priors in partially identified causal discove…

Compute & Architecture· 2026-11-13

Mr.Xu0 views0 discussions

Together Compiler Uses a Continual Agent Loop to Accelerate AI Inference

Together AI has introduced Together Compiler (TKCC), a continual compilation system for pr…

Agentic· 2026-11-13

Mr.Xu0 views0 discussions

Cleaning Agent History Can Backfire When the World Still Contains the Error

This study separates an agent's conversational history from the state of the external worl…