CHANNEL ARCHIVE // [166] ARTICLES // PAGE 3 OF 12
Artificial Intelligence
Research into artificial intelligence and intelligent systems.
CHANNEL // 01 [166]
Artificial Intelligence
Research into artificial intelligence and intelligent systems.
A 5G Jamming Benchmark Exposes a Blind Spot in Label-Free Detector Scores
Source published
- 5G
- Open RAN
- jamming detection
Examples improve Roman Urdu toxicity detection, but prompt-tuning gains remain modest
Source published
- Roman Urdu
- toxicity classification
- hate speech detection
When an AI experiment’s frozen protocol needs a bug fix
Source published
- pre-registration
- LLM-assisted research
- protocol repair
Teaching a live jamming detector with hindsight
Source published
- radio jamming
- O-RAN
- reinforcement learning
FP4Quant speeds up GPU communication, with a measurable accuracy trade-off
Source published
- FP4Quant
- four-bit quantization
- communication compression
Keeping AI agents within their assigned authority
Source published
- multi-agent systems
- large language models
- role drift
One AI can plan the program while another writes the code
Source published
- code generation
- large language models
- planning-coding separation
ATLAS steers AI agents’ hidden states to reduce coordination drift
Source published
- multi-agent AI
- ATLAS
- activation steering
A two-stage reasoning system helps AI choose Earth-observation evidence
Source published
- Earth observation
- geospatial question answering
- large language models
Teaching diffusion models to do more with fewer decoding steps
Source published
- discrete diffusion
- knowledge distillation
- score matching
Harness-S trains AI agents to retain lessons from specialized execution support
Source published
- AI agents
- agent harnesses
- reinforcement learning
Query-Matching Documents Can Crowd Out Evidence in RAG Retrieval
Source published
- retrieval-augmented generation
- corpus poisoning
- retrieval reliability
A learned token can hide unwanted AI behavior—but does not erase it
Source published
- artificial intelligence
- AI alignment
- inoculation prompting
AceAction maps the steps between game instructions and player actions
Source published
- AceAction-5M
- gameplay datasets
- vision-language-action models
A Checkable Proof of the Best Solution to One Optimization Benchmark
Source published
- quadratic optimization
- QPLIB
- set partitioning