archive
Every paper Pith has read. Search by title, abstract, or pith.
23781 papers in cs · page 2
-
Replaying early insights lifts accuracy in every LLM reasoning test
Stateful Reasoning via Insight Replay
-
IIQ metric scores AI integration from 0 to 1000
Intelligence Impact Quotient (IIQ): A Framework for Measuring Organizational AI Impact
-
LiSA adapts fixed AI guardrails over time by storing failures as reusable policy…
LiSA: Lifelong Safety Adaptation via Conservative Policy Induction
-
CP-OFDM yields lower ranging CRB than frequency-spread waveforms for large N
CP-OFDM Achieves Lower Ranging CRB Than Frequency-Spread Waveforms in the Large-Sample Regime
-
RL prompter lifts frozen LLM reasoning from 55% to 90%
Prompting Policies for Multi-step Reasoning and Tool-Use in Black-box LLMs with Iterative Distillation of Experience
-
The paper introduces a hybrid method for synthesizing policies in POMDPs by combining…
Synthesizing POMDP Policies: Sampling Meets Model-checking via Learning
-
I2P mixes peers randomly across countries
Geographic Patterns in I2P Peer Selection: An Empirical Network Topology Analysis
-
Generative prior plus sensor branches improves satellite rain estimates
A plug-and-play generative framework for multi-satellite precipitation estimation
-
3D scene edits yield consistent supervision for removal tasks
SceneForge: Structured World Supervision from 3D Interventions
-
3D solver models fluid-driven slip and tensile ruptures together
Three dimensional simulation of fluid-driven frictional and tensile ruptures on existing discontinuities
-
Diffusion models uncover semantic attacks on vehicle maps
Systematic Discovery of Semantic Attacks in Online Map Construction through Conditional Diffusion
-
Object clusters let 3D trajectories transfer across scenes
Analogical Trajectory Transfer
-
Dual-latent decoder balances fidelity and perception
Dual-Latent Collaborative Decoding for Fidelity-Perception Balanced Image Compression
-
Delta Forcing curbs drift in interactive video generation
Delta Forcing: Trust Region Steering for Interactive Autoregressive Video Generation
-
This paper proves tight bounds on how many agents an adversary can deactivate each round…
Semi-Synchronous Exploration in Dynamic Graphs
-
Exemplar partitioning matches top SAEs at 1000x lower cost
Exemplar Partitioning for Mechanistic Interpretability
-
NN radii converge almost surely under polynomial mixing
Nearest-Neighbor Radii under Dependent Sampling
-
Token module improves DDI predictions for unseen drugs
AIM-DDI: A Model-Agnostic Multimodal Integration Module for Drug-Drug Interaction Prediction
-
Dynamic Latent Routing beats supervised fine-tuning by 6.6 points
Dynamic Latent Routing
-
This paper proposes a gradient-based guidance framework that steers diffusion sampling…
Guided Diffusion Sampling for Precipitation Forecast Interventions
-
Top AI detectors lose most accuracy to paraphrased text
Paraphrasing Attack Resilience of Various AI-Generated Text Detection Methods
-
KAN network outperforms priors on hyperspectral-LiDAR fusion
Implicit spatial-frequency fusion of hyperspectral and lidar data via kolmogorov-arnold networks
-
Active learners raise NDCG@10 per call in PRP reranking
Active Learners as Efficient PRP Rerankers
-
Entangled QMARL agents near 0.85 win rate in CHSH game
Quantum Advantage in Multi Agent Reinforcement Learning
-
Masked contrastive pairs beat audio reconstruction
AudioMosaic: Contrastive Masked Audio Representation Learning
-
Endogenous trust cuts misinformation impact by 91% above threshold
Day-to-Day Traffic Network Modeling under Route-Guidance Misinformation: Endogenous Trust and Resilience in CAV Environments
-
Sparsifier keeps expected matching size with local k-edge budget
Stochastic Matching via Local Sparsification
-
Imagined future steps triple recall of distant memories
Thinking Ahead: Prospection-Guided Retrieval of Memory with Language Models
-
Univariate bicycle codes parametrize logical operators explicitly
Univariate Bicycle Quantum LDPC Codes: Explicit Logical Structure and Distance Bounds
-
Parallel BP on multiple bases cuts quantum LDPC errors up to 30%
Multiple-Bases Belief Propagation List Decoding for Quantum LDPC Codes
-
Mock agents hide real paths during multi-agent planning
Privacy Preserving Multi Agent Path Finding
-
One Rust visualization runs in Python
Pluot: Towards 'write once, run everywhere' visualization software
-
The paper presents a static data structure for leaf-to-ancestor minimum queries on rooted…
Fast Leaf-to-Ancestor Minimum Query in the Oracle Model
-
LLM facilitators nudge charity donations without raising group consensus
Real-Time Group Dynamics with LLM Facilitation: Evidence from a Charity Allocation Task
-
Six hours of data let a two-stage model beat larger ones on Wardaman
WARDEN: Endangered Indigenous Language Transcription and Translation with 6 Hours of Training Data
-
New logic QLL turns constraints into neural losses that match verifiers
Quantitative Linear Logic for Neuro-Symbolic Learning and Verification
-
No voice agent tops 0.5 on both accuracy and experience
EVA-Bench: A New End-to-end Framework for Evaluating Voice Agents
-
Valiant learnability equals poly-size query compression
What is Learnable in Valiant's Theory of the Learnable?
-
Agent weight updates cut token use 83% while raising accuracy
Good Agentic Friends Do Not Just Give Verbal Advice: They Can Update Your Weights
-
Rectification offset aligns mismatched meshes for video 4D animation
R-DMesh: Video-Guided 3D Animation via Rectified Dynamic Mesh Flow
-
Reachable sets turn HVAC coupling into real-time flexibility intervals
Reachable-Set Decomposition for Real-Time Aggregation of Multi-Zone HVAC Fleets
-
Hodge split isolates topology from geometry in neural field operators
Topology-Preserving Neural Operator Learning via Hodge Decomposition
-
Patch alignment lifts CLIP class-incremental learning
Unlocking Patch-Level Features for CLIP-Based Class-Incremental Learning
-
Quantum superposition memory processes long sequences linearly
QLAM: A Quantum Long-Attention Memory Approach to Long-Sequence Token Modeling
-
Balanced VQA data trains VLMs to handle 512K contexts after 128K training
Training Long-Context Vision-Language Models Effectively with Generalization Beyond 128K Context
-
Algebraic diagrams count sensitive regions in tree ensembles
Quantifying Sensitivity for Tree Ensembles: A symbolic and compositional approach
-
Finetuning makes models believe claims labeled false
Negation Neglect: When models fail to learn negations in training
-
Bootstrapping cuts prediction churn 40-54% in ML chemistry
Reducing cross-sample prediction churn in scientific machine learning
-
The paper tests frontier LLMs in agent scenarios where a short history of forced harmful…
History Anchors: How Prior Behavior Steers LLM Decisions Toward Unsafe Actions
-
Guidance sets safe speed for loiter UAV corridor reinsertion
Loiter UAV Reinsertion Guidance for Fixed-wing UAV Corridors