Lego Sketch: A Scalable Memory-augmented Neural Network for Sketching Data Streams

2025/05/26 by Yuan Feng, Yukun Cao, Feng, Yuan +7
Computer Science · #Data Stream Mining Techniques #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neural Networks and Applications #Time Series Analysis and Forecasting

paper · pdf · doi:10.48550/arxiv.2505.19561

openalex publication_date 2025/05/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Sketches, probabilistic structures for estimating item frequencies in infinite data streams with limited space, are widely used across various domains. Recent studies have shifted the focus from handcrafted sketches to neural sketches, leveraging memory-augmented neural networks (MANNs) to enhance the streaming compression capabilities and achieve better space-accuracy trade-offs.However, existing neural sketches struggle to scale across different data domains and space budgets due to inflexible MANN configurations. In this paper, we introduce a scalable MANN architecture that brings to life the \it Lego sketch, a novel sketch with superior scalability and accuracy. Much like assembling creations with modular Lego bricks, the Lego sketch dynamically coordinates multiple memory bricks to adapt to various space budgets and diverse data domains. Our theoretical analysis guarantees its high scalability and provides the first error bound for neural sketch. Furthermore, extensive experimental evaluations demonstrate that the Lego sketch exhibits superior space-accuracy trade-offs, outperforming existing handcrafted and neural sketches. Our code is available at https://github.com/FFY0/LegoSketch_ICML.

Citations

• Learning-Based Heavy Hitters and Flow Frequency Estimation in Streams
• Improved Frequency Estimation Algorithms with and without Predictions
• Experimental Analysis of Large-scale Learnable Vector Storage Compression
• Meta-Learning in Neural Networks: A Survey
• Meta-Learning Neural Bloom Filters
• Communication-efficient distributed SGD with Sketching
• Compressing Gradient Optimizers via Count-Sketches
• Count-Min: Optimal Estimation and Tight Error Bounds using Empirical Error Distributions
• Hash Embeddings for Efficient Word Representations
• Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry\n and Semantics
• Matching Networks for One Shot Learning
• Associative Long Short-Term Memory
• Neural Turing Machines
• Memory Networks
• Scaling Graph-based Semi Supervised Learning to Large Number of Labels\n Using Count-Min Sketch
• Finding repeated elements

Related

• Sketch-a-Net that Beats Humans
• Enabling My Robot To Play Pictionary : Recurrent Neural Networks For Sketch Recognition
• SketchINR: A First Look into Sketches as Implicit Neural Representations
• A Neural Representation of Sketch Drawings
• Higher-order Count Sketch: Dimensionality Reduction That Retains Efficient Tensor Operations