Skip to content

Research Ideas

Fresh research opportunities generated by AI from trending papers. Click on any idea to explore it in detail.

153 ideas match

Prompt-Relevant Historical Memory Retrieval and Entropy-Calibrated Layer-Head KV Compression

Replace fixed-rule cache eviction with prompt- and visual-relevance retrieval and entropy-calibrated per-head compression to reduce semantic drift over minute-long video rollouts.

MovieRoll: A Systematic Framework for Rolling Out Long-Form Videos
🟑Fairly easyπŸ’‘Moderate InnovationMultiple GPUsRelated work found#Generative media#Retrieval & memory

Quantized Attention, Denoising-Decoder Pipelining, and Sharded VAE Bottlenecks for Real-Time Streaming Video

Accelerate streaming video generation toward real-time playback by quantizing transformer weights, sharding decoder bottleneck attention across GPUs, and overlapping denoising with decoding.

MovieRoll: A Systematic Framework for Rolling Out Long-Form Videos
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsRelated work found#Efficiency#Generative media

Difficulty- and Saturation-Aware Adaptive Task-Level Recursion Budgeting

Dynamically allocate test-time exploration budgets using early rollout divergence and early-stopping checks so agents avoid wasting inference tokens on easy or saturated tasks.

Harness^2: Recursive Agent Harnessing for an Open World
🟑Fairly easyπŸ’‘Moderate InnovationSingle GPU or lessRelated work found#Agents#Efficiency

Uncertainty-Aware Proposal and Rollout Allocation for Continual Recursive Self-Improvement

Dynamically allocate proposal counts and evaluation rollouts based on memory confidence to prevent wasted inference compute on low-uncertainty harness defects.

Self-Improvement Memory Enables Continual Recursive Self-Improvement
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsRelated work found#Efficiency#Agents

Hierarchical Retrieval-Augmented Self-Improvement Memory for Extended Optimization Horizons

Retrieve relevant harness failure histories dynamically using semantic embedding search so recursive self-improvement scales across long continual adaptation sequences without overflowing context limits.

Self-Improvement Memory Enables Continual Recursive Self-Improvement
🟑Fairly easyπŸ’‘Moderate InnovationSingle GPU or lessNo close match found#Agents#Retrieval & memory

Invariant Relational Graph Networks for Cross-Sensor Array Generalization

A relational graph neural network that encodes normalized inter-sensor kinetic ratios to achieve generalization across individual sensor array variations and manufacturing batches.

Breath-Dx: A Time Series Classification Benchmark for Disease Screening with eNose
🟠MediumπŸ’‘Moderate InnovationSingle GPU or lessNo close match found#Training#AI for science

Lookup-Oriented Hardware Acceleration for Written Sparse Associative Tables

A specialized inference engine using CPU hash tables and near-memory lookups to eliminate GPU kernel overhead for sparse concept-retrieval architectures.

Weight-Space Programming Doesn't Need Transformers
🟠MediumπŸ”¬Significant ResearchSingle GPU or lessRelated work found#Efficiency

Differentiable Relaxation of Written Associative Memories for Few-Shot Fine-Tuning

A differentiable relaxation of discrete hash tables and concept weights that allows gradient descent to fine-tune human-readable written weights on low-data regimes.

Weight-Space Programming Doesn't Need Transformers
🟑Fairly easyπŸ’‘Moderate InnovationSingle GPU or lessNo close match found#Training#Efficiency

Recursive Associative-Memory Graph Traversal for Multi-Step Scientific Reasoning

A recursive memory traversal mechanism that chains sparse concept-fact lookups across multiple latent steps to solve multi-hop reasoning questions without dense generative models.

Weight-Space Programming Doesn't Need Transformers
🟑Fairly easyπŸ’‘Moderate InnovationSingle GPU or lessNo close match found#Reasoning#Efficiency

Feature-Parity Verification and Latency-Accuracy Co-Evaluation for Served Frontier VLAs

Verify numerical feature parity between fast serving engines and evaluation pipelines while logging observation age during asynchronous robotic control.

ScaleVLA: Scaling Vision-Language-Action Models to 552 Billion Parameters
🟑Fairly easyπŸ”§Minor TweakMultiple GPUsRelated work found#Efficiency#Data & evaluation

Dual-Rate Hierarchical Control with Closed-Loop Residual Correction for Contact-Rich Bimanual Manipulation

Combine asynchronous frontier vision-language feature updates with a high-frequency closed-loop residual controller to handle contact-sensitive alignment and observation staleness in bimanual manipulation.

ScaleVLA: Scaling Vision-Language-Action Models to 552 Billion Parameters
πŸ”΄HardπŸ”¬Significant ResearchLarge clusterRelated work found#Robotics#Agents

Cross-Scale Representation Distillation from 552B MoE VLMs into Single-GPU Student Policies

Distill causal representations from frontier mixture-of-experts vision-language models into compact single-GPU student backbones so physical robots achieve high manipulation accuracy without multi-GPU hardware.

ScaleVLA: Scaling Vision-Language-Action Models to 552 Billion Parameters
🟠MediumπŸ’‘Moderate InnovationLarge clusterRelated work found#Robotics#Efficiency

Hierarchical Edge-Cloud Architecture for Sub-Second Reactive Replanning

Pair slow cloud frontier models with lightweight local policies for sub-second reactive parameter adaptation during dynamic robotic contact tasks.

ForceHarness: Turning Force into Action and Physics for Frozen Frontier Models
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsRelated work found#Robotics#Efficiency

Dynamic Routing and Branch Dropout for Asymmetric Multi-Branch Transformers

Implement dynamic token-level branch routing and branch dropout to allow token-adaptive parallel compute and mitigate runtime imbalances during branch execution.

Multi-Branch Transformer
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsRelated work found#LLMs#Efficiency

Dual-Axis KV Sharing: Integrating Joint-KV with Cross-Layer and Prefix-Shared Attention

Combine branch-wise KV sharing with cross-layer KV caching to eliminate redundant key-value memory traffic across both network depth and parallel architectural branches.

Multi-Branch Transformer
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsRelated work found#LLMs#Efficiency

Empirical Production Serving Benchmarking of MLA-DSA Attention with IndexShare

A benchmark and serving engine that evaluates combined MLA, DSA, and IndexShare architectures under realistic dynamic batching and serving traffic on physical accelerators.

TeleChat to Xing4: Advancing Frontier Agentic Models on Domestic Chips and Frameworks
πŸ”΄HardπŸ”¬Significant ResearchMultiple GPUsNo close match found#Efficiency#LLMs

Joint Inter-DP and Intra-Sequence Layout Optimization for Multi-Document Sparse Attention

A method that unifies inter-data-parallel sample dispatch and intra-sequence query-key block folding to balance computational load during long-context sparse attention pretraining.

TeleChat to Xing4: Advancing Frontier Agentic Models on Domestic Chips and Frameworks
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsRelated work found#Efficiency#Training

Multi-Seed Stability Analysis and Distributed Multi-NPU Collective Kernel Benchmarking

Add multi-seed trajectory repeats and multi-device collective workloads to the evaluation protocol so benchmark rankings capture search variance and distributed communication bottlenecks.

AscendGYM: A Cost-Aware Kernel Optimization Benchmark for Real-World Workloads on Ascend NPUs
πŸ”΄HardπŸ’‘Moderate InnovationLarge clusterNo close match found#Data & evaluation#Efficiency

Roofline-Anchored Hardware Counter Feedback to Break the Kernel Optimization Ceiling

Feed automated hardware roofline headroom and memory-stall diagnostics into the optimization loop so coding agents identify architectural bottlenecks instead of guessing code edits.

AscendGYM: A Cost-Aware Kernel Optimization Benchmark for Real-World Workloads on Ascend NPUs
🟠MediumπŸ”¬Significant ResearchMultiple GPUsRelated work found#Agents#Efficiency

Execution-Time and Hardware-Cost-Aware Scoring Beyond Token Budgets in Agentic Kernel Optimization

Incorporate compilation, profiling, and hardware execution wall-clock time into the optimization score so agents are rewarded for fast turnaround rather than low token usage alone.

AscendGYM: A Cost-Aware Kernel Optimization Benchmark for Real-World Workloads on Ascend NPUs
🟑Fairly easyπŸ’‘Moderate InnovationSingle GPU or lessNo close match found#Data & evaluation#Efficiency

Multi-Shape Generalization and End-to-End Workload Evaluation for NPU Kernel Optimization

Extend kernel optimization evaluation to dynamic tensor shapes, fused multistage operators, and end-to-end model throughput so agents cannot overfit single static input configurations.

AscendGYM: A Cost-Aware Kernel Optimization Benchmark for Real-World Workloads on Ascend NPUs
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsRelated work found#Efficiency#Data & evaluation

Optimizing Ingestion Latency via Asynchronous Graph Indexing

A method that decouples graph construction from ingestion streams to reduce the heavy preprocessing overhead in document-heavy environments.

MIRIX-2.0: Multi-Agent Memory System
🟑Fairly easyπŸ”§Minor TweakSingle GPU or lessNo close match found#Efficiency#Retrieval & memory

Coordinate-Independent Relative Pose and Scale Representation

A method that replaces absolute coordinate regression with relative graph-based embeddings to eliminate distance-from-origin accumulation errors in long-sequence reconstructions.

Ο€^4: A Foundation Model for 4D Scene Geometry and Motion
πŸ”΄HardπŸ”¬Significant ResearchLarge clusterRelated work found#Vision#Training

Controlled Block-Level Comparison of Content-Gated Recurrent Depth Readers

Benchmark recurrent residual readers at block granularity with content-dependent gating and matched heads to isolate when fixed-size memory can rival explicit history attention.

Loopback Transformer: Towards a Unified View of Residual Looped Transformers
🟑Fairly easyπŸ’‘Moderate InnovationMultiple GPUsRelated work found#Training#Efficiency

Hybrid Sliding-Window and Fixed-State Depth Memory for Deeply Looped Transformers

Compress older layer updates into a fixed-size recurrent state while keeping recent updates explicit to cut memory traffic and latency in parameter-shared transformers.

Loopback Transformer: Towards a Unified View of Residual Looped Transformers
πŸ”΄HardπŸ”¬Significant ResearchMultiple GPUsNo close match found#Efficiency#Training

Modality-Dependence Reweighting and Interleaved Sparse Attention for Omni-Modal Audio-Visual-Speech Models

Generalize modality-dependence loss reweighting and interleaved sparse attention to audio-visual and streaming speech models so speculative decoding accelerates omni-modal generation.

MMFlash: Faster Speculative Decoding for Vision-Language Models
πŸ”΄HardπŸ”¬Significant ResearchMultiple GPUsNo close match found#Multimodal#Speech & audio

Single-Pass Visual Dependence Estimation to Eliminate Dual-Pass Target Scoring

Estimate token-level visual dependence from a single target forward pass via hidden-state shifts or attention statistics to eliminate the extra no-image scoring pass.

MMFlash: Faster Speculative Decoding for Vision-Language Models
🟑Fairly easyπŸ’‘Moderate InnovationMultiple GPUsNo close match found#Training#Efficiency

Intra-Block Causal Conditioning and Uncertainty-Triggered Visual Refetching for Visually Grounded Spans

Add lightweight intra-block causal conditioning and uncertainty-triggered visual cross-attention to parallel diffusion drafters so tokens within visually grounded spans condition on earlier predictions.

MMFlash: Faster Speculative Decoding for Vision-Language Models
πŸ”΄HardπŸ”¬Significant ResearchMultiple GPUsNo close match found#Multimodal#Efficiency

Concurrency- and Confidence-Adaptive Block-Size Scheduling for High-Batch VLM Serving

Adapt draft block lengths dynamically using serving batch size and predicted token acceptance confidence so speculative decoding avoids wasted verification compute under high concurrency.

MMFlash: Faster Speculative Decoding for Vision-Language Models
🟑Fairly easyπŸ”§Minor TweakSingle GPU or lessNo close match found#Efficiency#Multimodal

Query-Adaptive Video Context Compression and Frame-Level Visual Dependence for Long-Video Speculative Drafting

Select video draft tokens dynamically by question relevance and score visual dependence per frame so long-video drafters focus on informative moments instead of uniform strides.

MMFlash: Faster Speculative Decoding for Vision-Language Models
🟠MediumπŸ’‘Moderate InnovationMultiple GPUsNo close match found#Multimodal#Efficiency