Optimized Product Quantization

By Tiezheng Ge, Kaiming He, Qifa Ke. 2014

506 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Natural Language Processing

Citations in the corpus

Billion-Scale Similarity Search with GPUs (2017, methodology): “The optimized product quantization or OPQ [17] is a linear transformation on the input vectors that improves the accuracy of the product quantization; it can be applied as a pre-processing.”
Accelerating Very Deep Convolutional Networks for Classification and Detection (2015, background): “for various data reconstruction problems [34], [35], [36].”
Get Out of My Lab: Large-scale, Real-Time Visual-Inertial Localization (2015, background): “Decomposing the descriptor space such that each component has a similar variance reduces the quantization error of product quantization [17].”
Cache locality is not enough: High-Performance Nearest Neighbor Search with Product Quantization Fast Scan (2015, unknown): “Overall, PQ 8 × 8 provides the best performance tradeoﬀ, and is the most commonly used conﬁguration in the literature [14, 27, 4, 10, 21].”
Exploiting the Structure: Stochastic Gradient Methods Using Raw Clusters (2016, methodology): “[11] proposed to use approximate nearest neighbor algorithms such as LSH [2, 3] and product quantization [10, 12], and we use LSH in our experiments.”
Locally Optimized Product Quantization for Approximate Nearest Neighbor Search (2014, methodology): “We compare against three of the methods discussed in section 4.3, all using a single index on a coarse quantizer and PQ-encoded residuals, with any optimization being global.”
24/7 Place Recognition by View Synthesis (2018, unknown): “descriptors using product quantization (PQ) [21], [56], [57].”
Phrase Retrieval Learns Passage Retrieval, Too (2021, methodology): “See Ge et al. (2013) for more details on OPQ.”
Hashing for Distributed Data (2015, methodology): “A variety of ANN methods, such as iterative quantization (ITQ) (Gong et al., 2013), product quantization (Jegou et al., 2011), Cartesian k-means (Norouzi & Fleet, 2013) can be formulated within a unified framework with minimizing quantization distortion as objective (Ge et al., 2014).”
Towards Neural Theorem Proving at Scale (2018, methodology): “Exact methods are rarely more efficient than brute-force linear scan methods when the dimensionality is high (Ge et al., 2014; Malkov and Yashunin, 2016).”
Temporal Localization of Moments in Video Collections with Natural Language (2019, methodology): “As current efficient indexing techniques rely on approximating the Euclidean distance between descriptors [11, 13, 18], they cannot be readily plugged into video moment retrieval systems that rely on computing similarities using, often complicated, neural network architectures [4, 10, 23, 24].”
Sketch-based manga retrieval using manga109 dataset (2015, methodology): “The two main branches of ANN methods include productquantization-based methods [18, 26] and hashing-based methods [36, 68, 69, 73].”
Multiscale Quantization for Fast Similarity Search (2017, methodology): “A learned rotation is often applied to the residuals prior to product quantization to further reduce quantization error [11].”
Approximate Nearest Neighbor Search on High Dimensional Data - Experiments, Analyses, and Improvement (v1.0) (2016, unknown): “4.6 Optimal Product Quantization (OPQ) Vector quantization (VQ) [21] is a popular and successful method for ANN search, which could be used to build inverted index for non-exhaustive search.”
Efficient Indexing of Billion-Scale Datasets of Deep Descriptors (2016, methodology): “[9] improved the performance of Multi-D-ADC by replacing Product Quantization with Optimized Product Quantization for both indexing and compression (hence the name of their system OMulti-DOADC).”
Improving Bilayer Product Quantization for Billion-Scale Approximate Nearest Neighbors in High Dimensions (2014, methodology): “The system [5] forms this codebook via joint K-means clustering, while systems [1, 2] use product codebooks (based on product quantization [4]) for cells centroid construction.”
Milvus: A Purpose-Built Vector Data Management System (2021, methodology): “Existing works on vector data management mainly focus on vector similarity search [14, 20, 22, 33, 35, 39, 45, 46, 48, 49, 57, 65, 68], but they cannot meet the above requirements due to poor performance (on large-scale and dynamic vector data) and limited functionalities (e.g., not being capable…”
Accelerating Large-Scale Inference with Anisotropic Vector Quantization (2019, methodology): “Several techniques have been proposed in the literature based on hashing, graph search, or quantization to solve the approximate maximum inner product search problem efficiently, and the quantization based techniques have shown strong performance (Ge et al., 2014; Babenko & Lempitsky, 2014; Johnson ”