Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

169,096 papers · 148 categories

Trend · papers per month

4081121161 · Jun 202019922001200920182026
48 results for spatial memory

Generative model learns spatial memory for predicting agent's future in complex environments.

problem Training generative and temporal models in partially observed, 3D environments is challenging.
method Action-conditioned generative model with non-parametric spatial memory and state-space model for dynamics.
result Scalable architecture capable of coherent predictions over hundreds of time steps in 2D and 3D environments.

New neural network model learns to generalize spatial knowledge from memories.

problem Understanding how the brain generalizes structural knowledge from past experiences.
method Inspired by hippocampal-entorhinal system, proposes separating entity and structural representations.
result Artificial neural networks can learn structural knowledge and generalize it to new situations.

New CTRW model with memory explains long-term return autocorrelation.

problem Explaining long-term autocorrelation in financial returns.
method Proposed a Directed Continuous-Time Random Walk (CTRW) model with memory, considering only positive jumps and dependence on previous jumps.
result Bid-ask bounce explains only a small fraction of the long-term autocorrelation in financial returns.

MIP framework improves urban flow prediction by adapting to distribution shifts.

problem Distribution shifts in urban flow data make prediction models unreliable.
method Memory-enhanced Invariant Prompt learning with learnable memory bank.
result MIP ensures robust predictions by focusing on invariant features.

Product Kanerva Machines dynamically combine smaller models for better memory organization.

problem Limited organization in the Kanerva Machine.
method Introducing Product Kanerva Machines that dynamically combine multiple smaller Kanerva Machines.
result Product Kanerva Machines can discover spatial tunings that approximately factorize simple images by object.

A new method for faster spatial modeling on exascale computers.

problem Scalable, memory-efficient machine learning for spatially distributed data.
method Partitioned Sparse Variational Gaussian Process (PSVGP) with decentralized communication.
result Improved spatial predictions and better model fit with minimal overhead.

Novel graph model forecasts urban traffic with reduced spatial complexity.

problem Challenges in traffic forecasting due to spatio-temporal complexity, especially in urban environments.
method MW-TGC network model that combines spatial and temporal dependencies using multi-weighted adjacency matrices and graph convolution operations.
result MW-TGC network outperforms other models in urban-core and urban-mix sites, reducing variance in heterogeneous environments.

New method speeds up sparse Gaussian processes for large datasets.

problem Efficiently modeling large datasets with many inducing variables.
method Projecting a GP onto B-spline basis functions for sparse linear algebra.
result Efficiently models fast-varying spatial phenomena with tens of thousands of inducing variables.

We extend the theory of combinatorial link Floer homology to a class of oriented spatial graphs called transverse spatial graphs. To do this, we define the notion of a grid diagram representing a transverse spatial graph, which we call a graph grid diagram. We prove that two graph grid diagrams representing the same tr…

2015-06-15abs ↗pdf ↗

A deep historical LSTM model improves tennis shot recognition from RGB videos.

problem Improving action recognition from RGB video input for sports analysis.
method Convolutional Neural Network (CNN) and Weighted Long Short-Term Memory (LSTM) for 3D tennis shot recognition.
result The method achieves better performance than state-of-the-art baselines for tennis shot recognition.

PipeMare enables efficient DNN training with minimal memory and pipeline sacrifices.

problem Sacrificing hardware efficiency to maintain statistical efficiency in pipeline parallel DNN training.
method PipeMare is a simple yet robust training method that tolerates asynchronous updates during pipeline parallelism without sacrificing pipeline utilization or memory.
result PipeMare achieves up to 2.7x less memory usage or 4.3x higher pipeline utilization compared to state-of-the-art synchronous PP training techniques.

New neural networks model for spatio-temporal data.

problem Building a mapping from spatially encoded time series covariates to real-valued response data.
method Proposed two novel extensions of Functional Neural Network (FNN) for spatio-temporal regression.
result Demonstrated effectiveness in handling varying spatial correlations through comprehensive simulation studies.

Stacked LSTM improves weather forecasting accuracy by incorporating spatial information.

problem Improving temperature prediction accuracy in weather forecasting.
method 2-layer spatio-temporal stacked LSTM model with independent LSTM models per location in the first layer and combined hidden states in the second layer.
result The stacked LSTM model outperforms single LSTM models in most cases by utilizing spatial information.

Study compares tessellation strategies for taxi demand-supply forecasting models.

problem Improving taxi demand-supply forecasting using neural networks.
method Compared Voronoi tessellation and Geohash tessellation for LSTM models.
result Variable-sized polygon tessellation yields superior performance in LSTM models.

New framework models complex spatial data with basis functions and graphical vectors.

problem Modeling highly-multivariate spatial processes with varying resolutions.
method Extends graphical lasso to multivariate Gaussian processes with independent graphical vectors at different resolutions, using an orthogonal basis and fusion penalty.
result Linear complexity and parsimonious conditional independence structure in multilevel graphical model.

Paper predicts spatial variation data from few samples using tensor methods.

problem Predict spatial variation data from limited samples in high-dimensional data.
method Bayesian tensor completion exploiting hidden low-rank property.
result Predicts spatial variation data efficiently from few samples.

Graph Element Networks adaptively model spatial processes without prior graph structure.

problem Modeling spatial processes without a priori graphical structure.
method Assign nodes to spatial locations, optimize their connectivity, and use GNNs as a computational substrate.
result Optimized GNN nodes focus on complex parts of the space, allowing generalization and varying precision.

Vecchia approximations provide the best accuracy-runtime trade-off for Gaussian process approximations.

problem High computational cost of Gaussian processes for large data sets.
method Systematic comparison of different Gaussian process approximations.
result Vecchia approximations consistently provide the best accuracy-runtime trade-off.

Modeling the distribution of natural images is challenging, partly because of strong statistical dependencies which can extend over hundreds of pixels. Recurrent neural networks have been successful in capturing long-range dependencies in a number of problems but only recently have found their way into generative image…

2015-06-10abs ↗pdf ↗

New training algorithm enhances SNNs for temporal signal processing.

problem Lack of robust training algorithms for large-scale SNNs.
method Formulated SNN as IIR filters, proposed training algorithm for optimal synapse filter kernels and weights.
result Model and training algorithm outperform state-of-the-art approaches in accuracy.

A new tensor decomposition method for fMRI data captures both spatial and temporal variability.

problem Challenges in modeling shared and subject-specific structure in multisubject spatiotemporal data, especially in neuroimaging.
method Introduces a spatiotemporal variational tensor decomposition (ST-VTD) framework combining tensor factorization with structured priors for flexible representation of spatial and temporal dynamics.
result Significantly improves latent factor recovery in fMRI data compared to classical and probabilistic decomposition benchmarks.

MIM networks predict non-stationary spatiotemporal dynamics using differential signals.

problem Predicting non-stationary spatiotemporal processes with high-order variations.
method Memory In Memory (MIM) networks with cascaded memory modules.
result Achieved state-of-the-art results on four spatiotemporal prediction tasks.

Improved taxi demand-supply forecasts using graph-based LSTM.

problem Accurate taxi demand-supply forecasting with complex spatial and temporal patterns.
method Investigated impact of spatial partitioning techniques (Voronoi vs. Geohash) on LSTM network performance.
result GraphLSTM offers competitive performance against ConvLSTM, at lower complexity, across real-world data sets.

Study compares deep learning models for traffic forecasting, highlighting graph elements' impact.

problem Challenges in forecasting spatial-temporal traffic patterns.
method In-depth comparative study of four deep neural network models with different basic elements.
result Graph attention improves long-term predictions in traffic forecasting models.

CSTN predicts taxi demand between all regions, overcoming origin-only approaches.

problem Predicting taxi demand between all regions, not just origins.
method Contextualized Spatial-Temporal Network (CSTN) with LSC, TEC, and GCC modules.
result CSTN outperforms other methods in taxi origin-destination demand prediction.

New method improves neural decoding accuracy and reveals latent memory organization.

problem Improving neural decoding of temporal memory organization.
method Bayesian neural decoding using a diversity-encouraging latent representation learning method.
result Substantially higher accuracy in neural decoding and clear latent representation.

A new framework combines CNN and GRU for better structural damage detection.

problem Improving damage detection in structural engineering using machine learning.
method Hierarchical CNN and Gated Recurrent Unit (GRU) framework to model spatial and temporal relations.
result The proposed HCG framework significantly outperforms existing methods for structural damage detection.

COCO-GAN generates images by parts using spatial coordinates, achieving state-of-the-art quality.

problem Generating full images from partial observations due to computational constraints.
method COCO-GAN uses conditional coordination to generate images by parts based on spatial coordinates, with a generator and discriminator learning to justify realism.
result COCO-GAN produces state-of-the-art-quality full images during inference, enabling novel applications like beyond-boundary generation and panorama generation.