Search papers, labs, and topics across Lattice.
100 papers published across 11 labs.
Functional logic programs can be efficiently implemented in purely functional languages like Haskell, achieving performance gains over existing Curry compilers by using a novel monadic interface with memoization.
Get the best of both worlds: Linear-Core Surrogates offer the fast optimization of smooth losses and the statistical efficiency of margin-based losses, without sacrificing differentiability.
Tree-based RAG gets a major upgrade: $\Psi$-RAG's adaptive hierarchical index and multi-granular retrieval agent leapfrog existing methods on complex, cross-document reasoning tasks.
Decision trees and diffusion models are secretly doing the same thing: optimizing a shared objective called Global Trajectory Score Matching.
Jointly training the tokenizer and autoregressive model slashes ImageNet FID to 1.48, finally making end-to-end autoregressive image generation competitive.
Tree-based RAG gets a major upgrade: $\Psi$-RAG's adaptive hierarchical index and multi-granular retrieval agent leapfrog existing methods on complex, cross-document reasoning tasks.
Decision trees and diffusion models are secretly doing the same thing: optimizing a shared objective called Global Trajectory Score Matching.
Jointly training the tokenizer and autoregressive model slashes ImageNet FID to 1.48, finally making end-to-end autoregressive image generation competitive.
Instead of training separate video diffusion models for each multimodal task, UniVidX learns a single model that handles diverse pixel-aligned video generation problems.
Ditch the complex multimodal pre-training pipelines: GenLIP proves a simple language modeling objective can effectively align vision encoders with LLMs, achieving strong performance with less data.
LVLMs can maintain sharper visual focus during long-form generation by adding a lightweight, learnable memory module that bypasses attention dilution.
PINNs get a wavelet makeover, adaptively focusing on high-magnitude source regions and leaving vanilla methods in the dust on PDEs with extreme loss imbalances.
Signal processing practitioners gain a coherent roadmap for deploying sequential Gaussian Processes in real-world systems, bridging the gap between ML advances and practical application.
A single neural net can now solve 24 different multi-depot vehicle routing problems, thanks to a clever modulation technique that adapts to varying constraints.
Hyperbolic embeddings are powerful, but a fragmented ecosystem makes them hard to use—this framework finally puts them all in one place.
Ditch the Transformers: a cleverly designed all-MLP architecture, ITS-Mina, rivals state-of-the-art time series forecasting while slashing computational costs.
By combining Newton's method with adaptive gradient descent, this attractor FCM sidesteps premature convergence, offering a more robust approach to learning in complex cognitive maps.
Forget chasing bigger GPUs – the future of AI inference could be literally baked into the hardware itself, unlocking 1000x gains in energy and speed.
Standard GNNs can't cut it for solving linear SDPs, but a carefully designed architecture that mimics first-order solver updates can learn to predict solutions and dramatically accelerate convergence.
Transformers, typically considered inefficient for spin system sampling, can now outperform CNN-based samplers by generating groups of spins, unlocking larger system sizes and higher effective sample sizes.
Managing thousands of LEO satellites just got easier: a novel graph learning approach slashes network management overhead while boosting forecasting accuracy.
Hyperbolic embeddings and denoising diffusion can significantly boost few-shot learning on graphs, outperforming existing Euclidean-based methods.
Stop wasting compute on uninformative node types: TypeBandit intelligently allocates sampling resources in heterogeneous graphs, boosting attribute completion accuracy without architectural changes.
Enterprise AI doesn't have to be a latency nightmare: this pattern language offers a blueprint for integrating VLAs with deterministic control loops.
Despite its simplicity, mean pooling works surprisingly well because modern text encoders concentrate token embeddings, preserving crucial information about their distribution.
Forget turn-based interactions: MiniCPM-o 4.5 lets you build AI that sees, hears, speaks, and *reacts* in real-time, all on a device with only 12GB of RAM.
Sender-anonymity in quantum secret sharing is now possible, thanks to a clever combination of permutation-invariant codes and anonymous quantum transmission.
Angular apps are riddled with hidden design flaws: this study surfaces 11 common "code smells" and shows how to automatically sniff them out.
Stop costly cross-chain NFT migrations before they start: a new feature-centric methodology predicts which NFT functionalities will break when moving between blockchains like Ethereum and Solana.
Functional logic programs can be efficiently implemented in purely functional languages like Haskell, achieving performance gains over existing Curry compilers by using a novel monadic interface with memoization.
CNNs are surprisingly fragile to even single-pixel shifts, but strategically placed global average pooling can fix this with a 98% parameter reduction and no accuracy loss.
CNN classifiers don't just select from cleaned features, they actively cancel out shared background information via destructive interference, rewriting our understanding of how these networks actually "see".
Achieve high-fidelity 3D rendering from sparse, unconstrained real-world images by intelligently synthesizing novel views with diffusion models and Gaussian replication.
Forget fully connected relation graphs: CasLayout's sparse relation modeling unlocks enhanced controllability and realism in 3D indoor scene synthesis.
Simple, artist-friendly quad meshes can now be automatically generated on 3D shapes using a diffusion model trained on a continuous surface representation, sidestepping the complexity of discrete mesh optimization.
Achieve up to 2.5X faster video object removal by focusing DiT computations only on the essential tokens dictated by the mask.
Unlock the next level of robotic dexterity: this framework lets you co-design robotic hands by optimizing everything from palm structure to fingertip surface curvature.
Frustrated with clunky architecture simulators? Akita offers a breath of fresh air with its focus on developer experience, promising faster prototyping and experimentation.
NeuroRing achieves faster-than-real-time execution of a full-scale cortical microcircuit simulation on FPGAs, proving that scalable, energy-efficient SNN hardware is within reach.
Cerebras CS-3 can deliver 100x speedups over CPU for sparse matrix multiplication at 90% sparsity, but surprisingly, becomes *slower* than CPU beyond 99% sparsity.
Schedulers can boost throughput by 12% on chiplet-based systems simply by treating spatial locality as a first-class objective, even if it means sacrificing work-conservation.
Balancing processor utilization and Quality-of-Service in mixed-criticality systems just got easier with AnTi-MiCS and MulTi-MiCS, which automatically determine optimal low WCETs and improve QoS by up to 30%.
HBM-PIM can achieve impressive matrix multiplication throughput (14.9 GFLOP/s) using a novel reduction-free outer-product dataflow, even without native reduction support.
Forget waiting – this new CIM architecture slashes LLM weight update latency by up to 87%, unlocking faster prefill and decoding.
Ternary LLMs can achieve impressive throughput and energy efficiency on edge devices, thanks to VitaLLM's co-designed hardware acceleration that overcomes workload imbalance and data dependency challenges.
Ditch the costly sampling: Noise2Map turns diffusion models into fast, end-to-end semantic segmentation and change detection machines by directly predicting maps from noise.
By explicitly modeling uncertainty in hypergraph refinement, UHR-Net achieves more accurate segmentation of challenging lesions in medical images.
Machine learning can turn sparse simulation data into a complete phase diagram for collective motion models, revealing nuanced phase boundaries.
Discovering reusable, semantic "Action Motifs" from human movement data unlocks significant gains in action recognition, motion prediction, and interpolation.
Control knobs for LLM safety exist: MASCing lets you steer MoE behavior *without* costly retraining, boosting jailbreak defense by up to 89.2% and adult content generation control by up to 93.0%.
Hyperspherical latent spaces unlock better 3D scene understanding from vision transformers, especially when bandwidth is constrained.
Forget storing full task-specific models – Auto-FlexSwitch compresses the knowledge into tiny, dynamically assembled task vectors, slashing storage costs without sacrificing accuracy.
By fusing Bayesian neural networks with Kalman filtering, this work achieves more accurate and robust UAV state estimation than traditional methods in noisy, sparse sensing environments.
Sparse autoencoders, despite their popularity for extracting interpretable features, often fail to capture the underlying manifold structure of concepts, instead fragmenting them across multiple, diluted features.
Pinpointing the root cause of transformer failures just got a whole lot easier: DEFault++ can detect, categorize, and diagnose faults with high accuracy, even down to specific mechanisms.
A single KL identity unlocks a surprisingly simple and unified derivation of core results for exponential families, streamlining the theoretical foundations of variational inference, entropy-regularized RL, and RLHF.
Combining diverse AI prediction tools as a Mixture of Experts slashes variance in semi-supervised inference, outperforming standard Prediction-Powered Inference.
Modular architectures in continual learning only matter when representational dimensionality is low, revealing that dimensionality acts as a key control knob for the benefits of structural separation.
TwinGate stops jailbreaks by tracking malicious intent across anonymized, interleaved queries with minimal overhead, something previous defenses couldn't do.
Ditch the encoder-decoder: LPWTNet's closed-form Laplacian pyramid decomposition offers efficient inference for statistical channel fingerprint construction in massive MIMO systems.
Forget unreliable forecasts: CircuITS offers structurally guaranteed valid joint distributions for irregular multivariate time series, outperforming existing methods in joint and marginal density estimation.
Ditch the hash: training-free Hyper-Dimensional Fingerprints (HDF) unlock molecular representations with superior structural fidelity and property prediction compared to conventional methods, even at low dimensions.
Self-supervised encoders implicitly perform soft clustering on a "predictive manifold" in probability space, and this geometric perspective yields a practical Gaussian regularizer (SIGReg) competitive with variational IB.
Get the best of both worlds: Linear-Core Surrogates offer the fast optimization of smooth losses and the statistical efficiency of margin-based losses, without sacrificing differentiability.
Feature-level contrastive learning with dynamic masking unlocks superior performance on tabular remote sensing data, even when labels are scarce.
Forget noisy starts – ABC diffusion models leverage the inherent structure of continuous processes, generating future states from already-close previous states for more realistic dynamics.
LLMs can prune noisy edges in EEG graphs, leading to more accurate and interpretable seizure detection.
Get 4x-10x smaller LoRA models for free with a simple post-processing step that doesn't hurt performance.
Discovering new molecules and materials just got 10x cheaper, thanks to a hybrid AI method that blends generative models with physics-based search.
You can now get real-time (825 FPS) crack detection on UAVs without sacrificing accuracy, thanks to a new attention-enhanced lightweight CNN.
Turns out, arranging words to minimize syntactic dependency distance in sentences with star-like structures is easier than we thought, suggesting other factors drive word order.
Transformer-based models aren't always the only answer: SVMs offer a surprisingly competitive and efficient alternative for sentiment analysis, even when contextual understanding is key.
The best cryptographic Boolean networks aren't defined by a single architecture, but by surprisingly sparse and synergistic combinations of structural constraints.
Lattice reduction, long a dark art, can now be understood as minimizing variance in a Gram-Schmidt profile, leading to new, efficient heuristics.
Turns out, the best template for documenting architectural decisions depends on whether you value conciseness (Nygard) or structural detail (MADR).
Night photography can now look stunningly realistic, thanks to a new rendering technique that beats existing methods on perceptual quality and color accuracy.
Despite advances in deep learning, manufacturing-focused 3D reconstruction still struggles with reflective surfaces and dynamic environments, highlighting the need for robust hybrid systems.
Controllable 3D generation takes a leap forward with 3D-ReGen, a framework that leverages an initial 3D shape for tasks like enhancement and editing, outperforming existing methods.
NeRFs get a boost in video reconstruction quality by explicitly modeling inter- and intra-ray similarities with a novel transformer architecture.
Ditch the post-capture processing bottleneck: FUN achieves real-time hyperspectral object detection by jointly learning reconstruction and detection in a single, efficient network.
Ditch the clunky external tools: VeraRetouch slashes model size and unlocks end-to-end training for photo retouching with a fully differentiable architecture.
Sparsifying attention maps based on channel-wise correlations unlocks state-of-the-art performance in hyperspectral image super-resolution while maintaining competitive efficiency.
Continuous-depth transformers, augmented with physics-informed loss, can significantly improve short-term weather forecasting, suggesting a promising path for hybrid physics-aware AI models.
By representing deformable linear objects as a chain of relative rotations, RopeDreamer achieves state-of-the-art prediction accuracy and topological consistency in long-horizon manipulation tasks.
Origami tentacles that deterministically coil and stochastically entangle offer a surprisingly simple and robust solution for universal robotic gripping.
Training LLMs on ultra-long contexts just got a whole lot easier: AutoSP automates sequence parallelism and activation checkpointing, boosting context length by up to 2.7x with negligible throughput cost.
Even with adversarial network changes and only local signals, surprisingly simple distributed algorithms can enable dynamic networks to self-organize and adapt to changing environmental goals.
Slash MoE serving costs by two-thirds with FaaSMoE, a serverless architecture that dynamically scales experts on demand.
Dense matrix multiplication accelerators can surprisingly outperform dedicated sparse accelerators for sparse neural networks, offering better area and energy efficiency.
A holistic, industrial-grade V&V loop promises to accelerate and de-risk RISC-V chip design by integrating RTL validation, FPGA-based system-level testing, and continuous integration.
Emulating massive multi-core systems just got easier: EMiX lets you scale RISC-V emulation across multiple FPGAs without rewriting your RTL.
Skip the SCF convergence grind: a physically-constrained equivariant neural net slashes the number of iterations needed by up to 81% while also predicting accurate molecular properties in a single shot.
GNNs can predict core-electron binding energies in organic molecules with surprising accuracy (0.33 eV error), offering a computationally efficient alternative to expensive quantum chemistry calculations.
Transfer learning can unlock scalable emission control across diverse waste incineration plants by learning transferable system-level structures that capture physical constraints, operating-regime heterogeneity, and carbon-pollutant coupling.
Injecting review semantics into collaborative filtering via adaptive gating and contrastive learning substantially boosts top-N recommendation accuracy, outperforming existing review-aware methods.
Encrypting images no longer has to mean sacrificing accuracy in clothing classification tasks, thanks to a clever Vision Transformer approach.
Quantum circuits can boost classical U-Net performance in remote sensing image segmentation, even with shallow, parameter-efficient designs.
LLMs struggle with structured 2D tasks when inputs are serialized into 1D, revealing a surprising performance gap compared to vision-augmented models that directly process the 2D layout.
Hybrid-thinking LLMs can be dramatically improved by simply separating the feed-forward pathways for reasoning and non-reasoning modes, leading to less leakage and better accuracy.
Shrinking diffusion LLMs by distilling across different architectures can yield surprisingly strong performance, even boosting code generation scores by 16 points on HumanEval.
Forget coarse sequence-level hacks: LenVM lets you precisely dial in token generation length, boosting a 7B model's length accuracy from 30.9 to 64.8 and crushing closed-source rivals.
Ditch softmax attention for sigmoid: it unlocks 25% better cell-type separation, 10% faster training, and rock-solid stability for biological foundation models.
Forget brute-force scaling: smarter tile and tensor mapping on 3D-stacked chips could unlock massive LLM inference gains.
Edge LLM inference gets a serious speed boost: DUAL-BLADE's dual-path KV cache slashes latency by up to 42% and doubles SSD utilization.
Training a 1024-node SOM on a billion-sample dataset in just over 6 minutes shatters previous scalability limits, thanks to a novel framework that leverages multi-GPU execution, out-of-memory streaming, and flexible topologies.
