PyTorchKR Papers

최신 논문

총 590편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1902.02970공식 코드 · Papers with Code

Binarized Knowledge Graph Embeddings

Tensor factorization has become an increasingly popular approach to knowledge graph completion(KGC), which is the task of automatically predicting missing facts in a knowledge graph. However, even with a simple model like CANDECOMP/PARAFAC(CP) tensor decomposition, KGC on existing knowledge graphs is impractical in resource-limited environments, as a large amount of memory is required to store parameters represented as 32-bit or 64-bit floating point numbers. This limitation is expected to become more stringent as existing knowledge graphs, which are already huge, keep steadily growing in scale. To reduce the memory requirement, we present a method for binarizing the parameters of the CP tensor decomposition by introducing a quantization function to the optimization problem. This method replaces floating point-valued parameters with binary ones after training, which drastically reduces the model size at run time. We investigate the trade-off between the quality and size of tensor factorization models for several KGC benchmark datasets. In our experiments, the proposed method successfully reduced the model size by more than an order of magnitude while maintaining the task performance. Moreover, a fast score computation technique can be developed with bitwise operations.

arXiv:1902.00613공식 코드 · Papers with Code

Understanding Composition of Word Embeddings via Tensor Decomposition

Word embedding is a powerful tool in natural language processing. In this paper we consider the problem of word embedding composition \--- given vector representations of two words, compute a vector for the entire phrase. We give a generative model that can capture specific syntactic relations between words. Under our model, we prove that the correlations between three words (measured by their PMI) form a tensor that has an approximate low rank Tucker decomposition. The result of the Tucker decomposition gives the word embeddings as well as a core tensor, which can be used to produce better compositions of the word embeddings. We also complement our theoretical results with experiments that verify our assumptions, and demonstrate the effectiveness of the new composition method.

arXiv:1902.00038공식 코드 · Papers with Code

BLOCK: Bilinear Superdiagonal Fusion for Visual Question Answering and Visual Relationship Detection

Multimodal representation learning is gaining more and more interest within the deep learning community. While bilinear models provide an interesting framework to find subtle combination of modalities, their number of parameters grows quadratically with the input dimensions, making their practical implementation within classical deep learning pipelines challenging. In this paper, we introduce BLOCK, a new multimodal fusion based on the block-superdiagonal tensor decomposition. It leverages the notion of block-term ranks, which generalizes both concepts of rank and mode ranks for tensors, already used for multimodal fusion. It allows to define new ways for optimizing the tradeoff between the expressiveness and complexity of the fusion model, and is able to represent very fine interactions between modalities while maintaining powerful mono-modal representations. We demonstrate the practical interest of our fusion model by using BLOCK for two challenging tasks: Visual Question Answering (VQA) and Visual Relationship Detection (VRD), where we design end-to-end learnable architectures for representing relevant interactions between modalities. Through extensive experiments, we show that BLOCK compares favorably with respect to state-of-the-art multimodal fusion models for both VQA and VRD tasks. Our code is available at https://github.com/Cadene/block.bootstrap.pytorch.

arXiv:1901.11167

A Generalized Language Model in Tensor Space

In the literature, tensors have been effectively used for capturing the context information in language models. However, the existing methods usually adopt relatively-low order tensors, which have limited expressive power in modeling language. Developing a higher-order tensor representation is challenging, in terms of deriving an effective solution and showing its generality. In this paper, we propose a language model named Tensor Space Language Model (TSLM), by utilizing tensor networks and tensor decomposition. In TSLM, we build a high-dimensional semantic space constructed by the tensor product of word vectors. Theoretically, we prove that such tensor representation is a generalization of the n-gram language model. We further show that this high-order tensor representation can be decomposed to a recursive calculation of conditional probability for language modeling. The experimental results on Penn Tree Bank (PTB) dataset and WikiText benchmark demonstrate the effectiveness of TSLM.

arXiv:1812.10637

Sparse Nonnegative CANDECOMP/PARAFAC Decomposition in Block Coordinate Descent Framework: A Comparison Study

Nonnegative CANDECOMP/PARAFAC (NCP) decomposition is an important tool to process nonnegative tensor. Sometimes, additional sparse regularization is needed to extract meaningful nonnegative and sparse components. Thus, an optimization method for NCP that can impose sparsity efficiently is required. In this paper, we construct NCP with sparse regularization (sparse NCP) by l1-norm. Several popular optimization methods in block coordinate descent framework are employed to solve the sparse NCP, all of which are deeply analyzed with mathematical solutions. We compare these methods by experiments on synthetic and real tensor data, both of which contain third-order and fourth-order cases. After comparison, the methods that have fast computation and high effectiveness to impose sparsity will be concluded. In addition, we proposed an accelerated method to compute the objective function and relative error of sparse NCP, which has significantly improved the computation of tensor decomposition especially for higher-order tensor.

arXiv:1812.01839

Few Sample Knowledge Distillation for Efficient Network Compression

Deep neural network compression techniques such as pruning and weight tensor decomposition usually require fine-tuning to recover the prediction accuracy when the compression ratio is high. However, conventional fine-tuning suffers from the requirement of a large training set and the time-consuming training procedure. This paper proposes a novel solution for knowledge distillation from label-free few samples to realize both data efficiency and training/processing efficiency. We treat the original network as "teacher-net" and the compressed network as "student-net". A 1x1 convolution layer is added at the end of each layer block of the student-net, and we fit the block-level outputs of the student-net to the teacher-net by estimating the parameters of the added layers. We prove that the added layer can be merged without adding extra parameters and computation cost during inference. Experiments on multiple datasets and network architectures verify the method's effectiveness on student-nets obtained by various network pruning and weight decomposition methods. Our method can recover student-net's accuracy to the same level as conventional fine-tuning methods in minutes while using only 1% label-free data of the full training data.

arXiv:1811.07503공식 코드 · Papers with Code

Compressing Recurrent Neural Networks with Tensor Ring for Action Recognition

Recurrent Neural Networks (RNNs) and their variants, such as Long-Short Term Memory (LSTM) networks, and Gated Recurrent Unit (GRU) networks, have achieved promising performance in sequential data modeling. The hidden layers in RNNs can be regarded as the memory units, which are helpful in storing information in sequential contexts. However, when dealing with high dimensional input data, such as video and text, the input-to-hidden linear transformation in RNNs brings high memory usage and huge computational cost. This makes the training of RNNs unscalable and difficult. To address this challenge, we propose a novel compact LSTM model, named as TR-LSTM, by utilizing the low-rank tensor ring decomposition (TRD) to reformulate the input-to-hidden transformation. Compared with other tensor decomposition methods, TR-LSTM is more stable. In addition, TR-LSTM can complete an end-to-end training and also provide a fundamental building block for RNNs in handling large input data. Experiments on real-world action recognition datasets have demonstrated the promising performance of the proposed TR-LSTM compared with the tensor train LSTM and other state-of-the-art competitors.

arXiv:1811.07428

The core consistency of a compressed tensor

Tensor decomposition on big data has attracted significant attention recently. Among the most popular methods is a class of algorithms that leverages compression in order to reduce the size of the tensor and potentially parallelize computations. A fundamental requirement for such methods to work properly is that the low-rank tensor structure is retained upon compression. In lieu of efficient and realistic means of computing and studying the effects of compression on the low rank of a tensor, we study the effects of compression on the core consistency; a widely used heuristic that has been used as a proxy for estimating that low rank. We provide theoretical analysis, where we identify sufficient conditions for the compression such that the core consistency is preserved, and we conduct extensive experiments that validate our analysis. Further, we explore popular compression schemes and how they affect the core consistency.

arXiv:1811.05863

Robust low-rank multilinear tensor approximation for a joint estimation of the multilinear rank and the loading matrices

In order to compute the best low-rank tensor approximation using the Multilinear Tensor Decomposition (MTD) model, it is essential to estimate the rank of the underlying multilinear tensor from the noisy observation tensor. In this paper, we propose a Robust MTD (R-MTD) method, which jointly estimates the multilinear rank and the loading matrices. Based on the low-rank property and an over-estimation of the core tensor, this joint estimation problem is solved by promoting (group) sparsity of the over-estimated core tensor. Group sparsity is promoted using mixed-norms. Then we establish a link between the mixed-norms and the nuclear norm, showing that mixed-norms are better candidates for a convex envelope of the rank. After several iterations of the Alternating Direction Method of Multipliers (ADMM), the Minimum Description Length (MDL) criterion computed from the eigenvalues of the unfolding matrices of the estimated core tensor is minimized in order to estimate the multilinear rank. The latter is then used to estimate more accurately the loading matrices. We further develop another R-MTD method, called R-OMTD, by imposing an orthonormality constraint on each loading matrix in order to decrease the computation complexity. A series of simulated noisy tensor and real-world data are used to show the effectiveness of the proposed methods compared with state-of-the-art methods.

arXiv:1811.05076공식 코드 · Papers with Code

Learning from Binary Multiway Data: Probabilistic Tensor Decomposition and its Statistical Optimality

We consider the problem of decomposing a higher-order tensor with binary entries. Such data problems arise frequently in applications such as neuroimaging, recommendation system, topic modeling, and sensor network localization. We propose a multilinear Bernoulli model, develop a rank-constrained likelihood-based estimation method, and obtain the theoretical accuracy guarantees. In contrast to continuous-valued problems, the binary tensor problem exhibits an interesting phase transition phenomenon according to the signal-to-noise ratio. The error bound for the parameter tensor estimation is established, and we show that the obtained rate is minimax optimal under the considered model. Furthermore, we develop an alternating optimization algorithm with convergence guarantees. The efficacy of our approach is demonstrated through both simulations and analyses of multiple data sets on the tasks of tensor completion and clustering.

arXiv:1811.00944

Spectral Methods from Tensor Networks

A tensor network is a diagram that specifies a way to "multiply" a collection of tensors together to produce another tensor (or matrix). Many existing algorithms for tensor problems (such as tensor decomposition and tensor PCA), although they are not presented this way, can be viewed as spectral methods on matrices built from simple tensor networks. In this work we leverage the full power of this abstraction to design new algorithms for certain continuous tensor decomposition problems. An important and challenging family of tensor problems comes from orbit recovery, a class of inference problems involving group actions (inspired by applications such as cryo-electron microscopy). Orbit recovery problems over finite groups can often be solved via standard tensor methods. However, for infinite groups, no general algorithms are known. We give a new spectral algorithm based on tensor networks for one such problem: continuous multi-reference alignment over the infinite group SO(2). Our algorithm extends to the more general heterogeneous case.

arXiv:1810.13098

Low-Rank Embedding of Kernels in Convolutional Neural Networks under Random Shuffling

Although the convolutional neural networks (CNNs) have become popular for various image processing and computer vision task recently, it remains a challenging problem to reduce the storage cost of the parameters for resource-limited platforms. In the previous studies, tensor decomposition (TD) has achieved promising compression performance by embedding the kernel of a convolutional layer into a low-rank subspace. However the employment of TD is naively on the kernel or its specified variants. Unlike the conventional approaches, this paper shows that the kernel can be embedded into more general or even random low-rank subspaces. We demonstrate this by compressing the convolutional layers via randomly-shuffled tensor decomposition (RsTD) for a standard classification task using CIFAR-10. In addition, we analyze how the spatial similarity of the training data influences the low-rank structure of the kernels. The experimental results show that the CNN can be significantly compressed even if the kernels are randomly shuffled. Furthermore, the RsTD-based method yields more stable classification accuracy than the conventional TD-based methods in a large range of compression ratios.

arXiv:1810.11896

Smoothed Analysis of Discrete Tensor Decomposition and Assemblies of Neurons

We analyze linear independence of rank one tensors produced by tensor powers of randomly perturbed vectors. This enables efficient decomposition of sums of high-order tensors. Our analysis builds upon [BCMV14] but allows for a wider range of perturbation models, including discrete ones. We give an application to recovering assemblies of neurons. Assemblies are large sets of neurons representing specific memories or concepts. The size of the intersection of two assemblies has been shown in experiments to represent the extent to which these memories co-occur or these concepts are related; the phenomenon is called association of assemblies. This suggests that an animal's memory is a complex web of associations, and poses the problem of recovering this representation from cognitive data. Motivated by this problem, we study the following more general question: Can we reconstruct the Venn diagram of a family of sets, given the sizes of their $\ell$-wise intersections? We show that as long as the family of sets is randomly perturbed, it is enough for the number of measurements to be polynomially larger than the number of nonempty regions of the Venn diagram to fully reconstruct the diagram.

arXiv:1810.10065공식 코드 · Papers with Code

Statistical mechanics of low-rank tensor decomposition

Often, large, high dimensional datasets collected across multiple modalities can be organized as a higher order tensor. Low-rank tensor decomposition then arises as a powerful and widely used tool to discover simple low dimensional structures underlying such data. However, we currently lack a theoretical understanding of the algorithmic behavior of low-rank tensor decompositions. We derive Bayesian approximate message passing (AMP) algorithms for recovering arbitrarily shaped low-rank tensors buried within noise, and we employ dynamic mean field theory to precisely characterize their performance. Our theory reveals the existence of phase transitions between easy, hard and impossible inference regimes, and displays an excellent match with simulations. Moreover, it reveals several qualitative surprises compared to the behavior of symmetric, cubic tensor decomposition. Finally, we compare our AMP algorithm to the most commonly used algorithm, alternating least squares (ALS), and demonstrate that AMP significantly outperforms ALS in the presence of noise.

arXiv:1810.08612

CNN inference acceleration using dictionary of centroids

It is well known that multiplication operations in convolutional layers of common CNNs consume a lot of time during inference stage. In this article we present a flexible method to decrease both computational complexity of convolutional layers in inference as well as amount of space to store them. The method is based on centroid filter quantization and outperforms approaches based on tensor decomposition by a large margin. We performed comparative analysis of the proposed method and series of CP tensor decomposition on ImageNet benchmark and found that our method provide almost 2.9 times better computational gain. Despite the simplicity of our method it cannot be applied directly in inference stage in modern frameworks, but could be useful for cases calculation flow could be changed, e.g. for CNN-chip designers.

arXiv:1810.07468공식 코드 · Papers with Code

Hierarchical Methods of Moments

Spectral methods of moments provide a powerful tool for learning the parameters of latent variable models. Despite their theoretical appeal, the applicability of these methods to real data is still limited due to a lack of robustness to model misspecification. In this paper we present a hierarchical approach to methods of moments to circumvent such limitations. Our method is based on replacing the tensor decomposition step used in previous algorithms with approximate joint diagonalization. Experiments on topic modeling show that our method outperforms previous tensor decomposition methods in terms of speed and model quality.

arXiv:1810.04754

Efficient Tensor Decomposition with Boolean Factors

Tensor decomposition has been extensively used as a tool for exploratory analysis. Motivated by neuroscience applications, we study tensor decomposition with Boolean factors. The resulting optimization problem is challenging due to the non-convex objective and the combinatorial constraints. We propose Binary Matching Pursuit (BMP), a novel generalization of the matching pursuit strategy to decompose the tensor efficiently. BMP iteratively searches for atoms in a greedy fashion. The greedy atom search step is solved efficiently via a MAXCUT-like boolean quadratic program. We prove that BMP is guaranteed to converge sublinearly to the optimal solution and recover the factors under mild identifiability conditions. Experiments demonstrate the superior performance of our method over baselines on synthetic and real datasets. We also showcase the application of BMP in quantifying neural interactions underlying high-resolution spatiotemporal ECoG recordings.

arXiv:1809.04110

Joint Embedding of Meta-Path and Meta-Graph for Heterogeneous Information Networks

Meta-graph is currently the most powerful tool for similarity search on heterogeneous information networks,where a meta-graph is a composition of meta-paths that captures the complex structural information. However, current relevance computing based on meta-graph only considers the complex structural information, but ignores its embedded meta-paths information. To address this problem, we proposeMEta-GrAph-based network embedding models, called MEGA and MEGA++, respectively. The MEGA model uses normalized relevance or similarity measures that are derived from a meta-graph and its embedded meta-paths between nodes simultaneously, and then leverages tensor decomposition method to perform node embedding. The MEGA++ further facilitates the use of coupled tensor-matrix decomposition method to obtain a joint embedding for nodes, which simultaneously considers the hidden relations of all meta information of a meta-graph.Extensive experiments on two real datasets demonstrate thatMEGA and MEGA++ are more effective than state-of-the-art approaches.

arXiv:1809.04096

Parallel Separable 3D Convolution for Video and Volumetric Data Understanding

For video and volumetric data understanding, 3D convolution layers are widely used in deep learning, however, at the cost of increasing computation and training time. Recent works seek to replace the 3D convolution layer with convolution blocks, e.g. structured combinations of 2D and 1D convolution layers. In this paper, we propose a novel convolution block, Parallel Separable 3D Convolution (PmSCn), which applies m parallel streams of n 2D and one 1D convolution layers along different dimensions. We first mathematically justify the need of parallel streams (Pm) to replace a single 3D convolution layer through tensor decomposition. Then we jointly replace consecutive 3D convolution layers, common in modern network architectures, with the multiple 2D convolution layers (Cn). Lastly, we empirically show that PmSCn is applicable to different backbone architectures, such as ResNet, DenseNet, and UNet, for different applications, such as video action recognition, MRI brain segmentation, and electron microscopy segmentation. In all three applications, we replace the 3D convolution layers in state-of-the art models with PmSCn and achieve around 14% improvement in test performance and 40% reduction in model size and on average.

arXiv:1809.03618공식 코드 · Papers with Code

Visualization of High-dimensional Scalar Functions Using Principal Parameterizations

Insightful visualization of multidimensional scalar fields, in particular parameter spaces, is key to many fields in computational science and engineering. We propose a principal component-based approach to visualize such fields that accurately reflects their sensitivity to input parameters. The method performs dimensionality reduction on the vast $L^2$ Hilbert space formed by all possible partial functions (i.e., those defined by fixing one or more input parameters to specific values), which are projected to low-dimensional parameterized manifolds such as 3D curves, surfaces, and ensembles thereof. Our mapping provides a direct geometrical and visual interpretation in terms of Sobol's celebrated method for variance-based sensitivity analysis. We furthermore contribute a practical realization of the proposed method by means of tensor decomposition, which enables accurate yet interactive integration and multilinear principal component analysis of high-dimensional models.