Hash function generation by neural network

Turčaník, Michal; Javurek, Martin

doi:10.1109/ntsp.2016.7747793

Cited by 14 publications

(10 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most notably, there has been work on learning locality-sensitive hash (LSH) functions to build Approximate Nearest Neighborhood (ANN) indexes. For example, [66,68,40] explore the use of neural networks as a hash function, whereas [69] even tries to preserve the order of the multi-dimensional input space. However, the general goal of LSH is to group similar items into buckets to support nearest neighborhood queries, usually involving learning approximate similarity measures in high-dimensional input space using some variant of hamming distances.…”

Section: Related Workmentioning

confidence: 99%

The Case for Learned Index Structures

Kraska

Beutel

et al. 2018

Proceedings of the 2018 International Conference on Management of Data

648

620

View full text Add to dashboard Cite

Indexes are models: a B-Tree-Index can be seen as a model to map a key to the position of a record within a sorted array, a Hash-Index as a model to map a key to a position of a record within an unsorted array, and a BitMap-Index as a model to indicate if a data record exists or not. In this exploratory research paper, we start from this premise and posit that all existing index structures can be replaced with other types of models, including deep-learning models, which we term learned indexes. We theoretically analyze under which conditions learned indexes outperform traditional index structures and describe the main challenges in designing learned index structures. Our initial results show that our learned indexes can have significant advantages over traditional indexes. More importantly, we believe that the idea of replacing core components of a data management system through learned models has far reaching implications for future systems designs and that this work provides just a glimpse of what might be possible.

show abstract

Section: Related Workmentioning

confidence: 99%

The Case for Learned Index Structures

Kraska

Beutel

et al. 2018

Proceedings of the 2018 International Conference on Management of Data

648

620

View full text Add to dashboard Cite

show abstract

“…However, order-preserving hashing is unsuitable for sorting since it does not provide fast enough training and inference times, and, to the best of our knowledge, there does not exist any sorting algorithm that uses order-preserving hashing for sorting. Similarly, localitysensitive hashing [54,56,57] can also not be used for sorting a single numeric value as we are concerned with sorting a single dimension rather than efficiently finding similar items in a multi-dimensional space. Finally, perfect hashing's objective is to avoid element collisions, which would initially seem an interesting choice to use for Learned Sort.…”

Section: Related Workmentioning

confidence: 99%

The Case for a Learned Sorting Algorithm

Kristo

Vaidya

Çetintemel

et al. 2020

Proceedings of the 2020 ACM SIGMOD International Conference on Management of Data

View full text Add to dashboard Cite

Sorting is one of the most fundamental algorithms in Computer Science and a common operation in databases not just for sorting query results but also as part of joins (i.e., sortmerge-join) or indexing. In this work, we introduce a new type of distribution sort that leverages a learned model of the empirical CDF of the data. Our algorithm uses a model to efficiently get an approximation of the scaled empirical CDF for each record key and map it to the corresponding position in the output array. We then apply a deterministic sorting algorithm that works well on nearly-sorted arrays (e.g., Insertion Sort) to establish a totally sorted order.We compared this algorithm against common sorting approaches and measured its performance for up to 1 billion normally-distributed double-precision keys. The results show that our approach yields an average 3.38× performance improvement over C++ STL sort, which is an optimized Quicksort hybrid, 1.49× improvement over sequential Radix Sort, and 5.54× improvement over a C++ implementation of Timsort, which is the default sorting function for Java and Python.

show abstract

“…The application of neural networks as a hash function is researched mainly in the context of security. Turčaník et al [32] initialize a neural network with random weights and biases. Lian et al [23] apply a chaotic map function to the weights and biases.…”

Section: Neural Network As Hash Functionsmentioning

confidence: 99%

“…This is possible because a key design feature for GNNs is that they should be permutation invariant or at least equivariant [13], which also provides the desired effect for this case. The application of neural networks for hash functions is mainly researched under a security perspective [23,32]. Their goal is to provide a one-way function that ensures big changes in the resulting hash, even when there are only small differences in the input.…”

Section: Introductionmentioning

confidence: 99%

Graph Summarization with Graph Neural Networks

Blasi¹,

Freudenreich²,

Horvath³

et al. 2022

Preprint

View full text Add to dashboard Cite

The goal of graph summarization is to represent large graphs in a structured and compact way. A graph summary based on equivalence classes preserves pre-defined features of a graph's vertex within a 𝑘-hop neighborhood such as the vertex labels and edge labels. Based on this neighborhood characteristics, the vertex is assigned to an equivalence class. The calculation of the assigned equivalence class must be a permutation invariant operation on the pre-defined features. This is achieved by sorting on the feature values, e. g., the edge labels, which is computationally expensive, and subsequently hashing the result. Graph Neural Networks (GNN) fulfill the permutation invariance requirement. We formulate the problem of graph summarization as a subgraph classification task on the root vertex of the 𝑘-hop neighborhood. We adapt different GNN architectures, both based on the popular message-passing protocol and alternative approaches, to perform the structural graph summarization task. We compare different GNNs with a standard multilayer perceptron (MLP) and Bloom filter as non-neural method.For our experiments, we consider four popular graph summary models on a large web graph. This resembles challenging multiclass vertex classification tasks with the numbers of classes ranging from 576 to multiple hundreds of thousands. Our results show that the performance of GNNs are close to each other. In three out of four experiments, the non-message-passing GraphMLP model outperforms the other GNNs. The performance of the standard MLP is extraordinary good, especially in the presence of many classes. Finally, the Bloom filter outperforms all neural architectures by a large margin, except for the dataset with the fewest number of 576 classes. This is an interesting result, since it shows an interaction effect between the number of classes for the summarization task and the chosen method.

show abstract

Hash function generation by neural network

Cited by 14 publications

References 2 publications

The Case for Learned Index Structures

The Case for Learned Index Structures

The Case for a Learned Sorting Algorithm

Graph Summarization with Graph Neural Networks

Contact Info

Product

Resources

About