United States

Kernel approximation is widely used to scale up kernel SVM training and prediction. However, the memory and computation costs of kernel approximation models are still too large if we want to deploy them on memory-limited devices such as mobile phones, smart watches and IoT devices. To address this challenge, we propose a novel memory and computation-efficient kernel SVM model by using both binary embedding and binary model coefficients. First, we propose an efficient way to generate compact binary embedding of the data which can preserve the kernel similarity. Second, we propose a simple but effective algorithm to learn a linear classification model with binary coefficients which can support different types of loss function and regularizer. Our algorithm can achieve better generalization accuracy than existing works on learning binary coefficients since we allow coefficient to be -1, 0 or 1 during the training stage and coefficient 0 can be removed during model inference. Moreover, we provide detailed analysis on the convergence of our algorithm and the inference complexity of our model. The analysis shows that the convergence to a local optimum is guaranteed and the inference complexity of our model is much lower than other competing methods. Our experimental results on five large real-world datasets have demonstrated that our proposed method can build accurate nonlinear SVM model with memory cost less than 30KB.

AAAI 2021

Memory and Computation-Efficient Kernel SVM via Binary Embedding and Ternary Coefficients

The purpose of the AAAI conference is to promote research in artificial intelligence (AI) and scientific exchange among AI researchers, practitioners, scientists, and engineers in affiliated disciplines. 

technical paper

The past two decades have seen the significant improvements of the scalability of practical model counters, which have been quite influential in many applications from artificial intelligence to formal verification.  While most of exact counters fall into two categories, search-based and compilation-based, Huang and Darwiche's remarkable observation ties these two categories: the trace of a search-based exact model counter corresponds to a Decision-DNNF formula.  Taking advantage of literal equivalences, this paper designs an efficient model counting technique such that its trace is a generalization of Decision-DNNF formula.  We first propose a generalization of Decision-DNNF, called CCDD, to capture literal equivalences, then show that CCDD supports model counting in linear time, and finally design a model counter, called ExactMC, whose trace corresponds to CCDD.  We perform an extensive experimental evaluation over a comprehensive set of benchmarks and conduct performance comparison of ExactMC vis-a-vis the state of the art counters, c2d, Dsharp, miniC2D, D4, ADDMC, and Ganak. Our empirical evaluation demonstrates ExactMC can solve 885 instances while the prior state of the art could solve only 843 instances, representing a significant improvement of 42 instances.

The Power of Literal Equivalence in Model Counting

Video generation models often operate under the assumption of fixed frame rates, which leads to suboptimal performance when it comes to handling flexible frame rates (e.g., increasing the frame rate of the more dynamic portion of the video as well as handling missing video frames). To resolve the restricted nature of existing video generation models' ability to handle arbitrary timesteps, we propose continuous-time video generation by combining neural ODE (Vid-ODE) with pixel-level video processing techniques. Using ODE-ConvGRU as an encoder, a convolutional version of the recently proposed neural ODE, which enables us to learn continuous-time dynamics, Vid-ODE can learn the spatio-temporal dynamics of input videos of flexible frame rates. The decoder integrates the learned dynamics function to synthesize video frames at any given timesteps, where the pixel-level composition technique is used to maintain the sharpness of individual frames. With extensive experiments on four real-world video datasets, we verify that the proposed Vid-ODE outperforms state-of-the-art approaches under various video generation settings, both within the trained time range (interpolation) and beyond the range (extrapolation). To the best of our knowledge, Vid-ODE is the first work successfully performing continuous-time video generation using real-world videos.

Vid-ODE: Continuous-Time Video Generation with Neural Ordinary Differential Equation

One-bit matrix completion is an important class of positive-unlabeled (PU) learning problems where the observations consist of only positive examples, e.g., in top-N recommender systems. For the first time, we show that 1-bit matrix completion can be formulated as the problem of recovering clean graph signals from noise-corrupted signals in hypergraphs. This makes it possible to enjoy recent advances in graph signal learning. Then, we propose the spectral graph matrix completion (SGMC) method, which can recover the underlying matrix in distributed systems by filtering the noisy data in the graph frequency domain. Meanwhile, it can provide micro- and macro-level explanations by following vertex-frequency analysis. To tackle the computational and memory issue of performing graph signal operations on large graphs, we construct a scalable Nystrom algorithm which can efficiently compute orthonormal eigenvectors. Furthermore, we also develop polynomial and sparse frequency filters to remedy the accuracy loss caused by the approximations. We demonstrate the effectiveness of our algorithms on top-N recommendation tasks, and the results on three large-scale real-world datasets show that SGMC can outperform state-of-the-art top-N recommendation algorithms in accuracy while only requiring a small fraction of training time compared to the baselines.

Scalable and Explainable 1-Bit Matrix Completion via Graph Signal Learning

Knowledge distillation extracts general knowledge from a pretrained teacher network and provides guidance to a target student network. Most studies manually tie intermediate features of the teacher and student, and transfer knowledge through predefined links. However, manual selection often constructs ineffective links that limit the improvement from the distillation. There has been an attempt to address the problem, but it is still challenging to identify effective links under practical scenarios. In this paper, we introduce an effective and efficient feature distillation method utilizing all the feature levels of the teacher without manually selecting the links. Specifically, our method utilizes an attention based meta network that learns relative similarities between features, and applies identified similarities to control distillation intensities of all possible pairs. As a result, our method determines competent links more efficiently than the previous approach and provides better performance on model compression and transfer learning tasks. Further qualitative analyses and ablative studies describe how our method contributes to better distillation.

Show, Attend and Distill: Knowledge Distillation via Attention-Based Feature Matching

In federated learning, models are learned from usersâ€™ data that are held private in their edge devices, by aggregating them in the service providerâ€™s â€œcloudâ€ to obtain a global model. Such global model is of great commercial value in, e.g., improving the customersâ€™ experience. In this paper we focus on two possible areas of improvement of the state of the art. First, we take the difference between user habits into account and propose a quadratic penalty-based formulation, for efficient learning of the global model that allows to personalize local models. Second, we address the latency issue associated with the heterogeneous training time on edge devices, by exploiting a hierarchical structure modeling communication not only between the cloud and edge devices, but also within the cloud. Specifically, we devise a tailored block coordinate descent-based computation scheme, accompanied with communication protocols for both the synchronous and asynchronous cloud settings. We characterize the theoretical convergence rate of the algorithm, and provide a variant that performs empirically better. We also prove that the asynchronous protocol, inspired by multi-agent consensus technique, has the potential for large gains in latency compared to a synchronous setting when the edge-device updates are intermittent. Finally, experimental results are provided that corroborate not only the theory, but also show that the system leads to faster convergence for personalized models on the edge devices, compared to the state of the art.

Federated Block Coordinate Descent Scheme for Learning Global and Personalized Models

Long-tailed class distributions are prevalent among the practical applications of object detection and instance segmentation. Prior work in long-tail instance segmentation addresses the imbalance of losses between rare and frequent categories by reducing the penalty for a model incorrectly predicting a rare class label. We demonstrate that the rare categories are heavily suppressed by correct background predictions, which reduces the probability for all foreground categories with equal weight. Due to the relative infrequency of rare categories, this leads to an imbalance that biases towards predicting more frequent categories. Based on this insight, we develop DropLoss -- a novel adaptive loss to compensate for this imbalance without a trade-off between rare and frequent categories. With this loss, we show state-of-the-art mAP across rare, common, and frequent categories on the LVIS dataset. Codes are available at https://github.com/timy90022/DropLoss.

DropLoss for Long-Tail Instance Segmentation

Automatic comment generation is a special and challenging task to verify the model ability on news content comprehension and language generation. Comments not only convey salient and interesting information in news articles, but also imply various and different reader characteristics which we treat as the essential clues for diversity. However, most of the comment generation approaches only focus on saliency information extraction, while the reader-aware factors implied by comments are neglected. To address this issue, we propose a unified reader-aware topic modeling and saliency information detection framework to enhance the quality of generated comments. For reader-aware topic modeling, we design a variational generative clustering algorithm for latent semantic learning and topic mining from reader comments. For saliency information detection, we introduce Bernoulli distribution estimating on news content to select saliency information. The obtained topic representations as well as the selected saliency information are incorporated into the decoder to generate diversified and informative comments. Experimental results on three datasets show that our framework outperforms existing baseline methods in terms of both automatic metrics and human evaluation. The potential ethical issues are also discussed in detail.

Generating Diversified Comments via Reader-Aware Topic Modeling and Saliency Detection

Combinatorial optimization has found applications in numerous fields, from aerospace to transportation planning and economics. The goal is to find an optimal solution among a finite set of possibilities. The well-known challenge one faces with combinatorial optimization is the state-space explosion problem: the number of possibilities grows exponentially with the problem size, which makes solving intractable for large problems. In the last years, deep reinforcement learning (DRL) has shown its promise for designing good heuristics dedicated to solve NP-hard combinatorial optimization problems. However, current approaches have an important shortcoming:  they only provide an approximate solution with no systematic ways to improve it or to prove optimality. In another context, constraint programming (CP) is a generic tool to solve combinatorial optimization problems. Based on a complete search procedure, it will always find the optimal solution if we allow an execution time large enough. A critical design choice, that makes CP non-trivial to use in practice, is the branching decision, directing how the search space is explored. In this work, we propose a general and hybrid approach, based on DRL and CP, for solving combinatorial optimization problems. The core of our approach is based on a dynamic programming formulation, that acts as a bridge between both techniques. We experimentally show that our solver is efficient to solve three challenging problems: the traveling salesman problem with time windows, the 4-moments portfolio optimization problem, and the 0-1 knapsack problem. Results obtained show that the framework introduced outperforms the stand-alone RL and CP solutions, while being competitive with industrial solvers.

Combining Reinforcement Learning and Constraint Programming for Combinatorial Optimization

State-of-the-art video object detection methods maintain a memory structure, either a sliding window or a memory queue, to enhance the current frame using attention mechanisms. However, we argue that these memory structures are not efficient or sufficient because of two implied operations: (1) concatenating all features in memory for enhancement, leading to a heavy computational cost; (2) frame-wise memory updating, preventing the memory from capturing more temporal information. In this paper, we propose a multi-level aggregation architecture via memory bank called MAMBA. Specifically, our memory bank employs two novel operations to eliminate disadvantages of existing methods: (1) light-weight key-set construction which can significantly reduce the computational cost; (2) fine-grained feature-wise updating strategy which enables our method to utilize knowledge from the whole video. To better enhance features from complementary levels, i.e., feature maps and proposals, we further propose a generalized enhancement operation (GEO) to aggregate multi-level features in a unified manner. We conduct extensive evaluations on the challenging ImageNetVID dataset. Compared with existing state-of-the-art methods, our method achieves superior performance in terms of both speed and accuracy. More remarkably, MAMBA achieves mAP of 83.7%/84.6% at 12.6/9.1 FPS with ResNet-101.

MAMBA: Multi-level Aggregation via Memory Bank for Video Object Detection

Recent work has increased the performance of Generative Adversarial Networks (GANs) by enforcing a consistency cost on the discriminator. We improve on this technique in several ways. We first show that consistency regularization can introduce artifacts into the GAN samples and explain how to fix this issue. We then propose several modifications to the consistency regularization procedure designed to improve its performance. We carry out extensive experiments quantifying the benefit of our improvements. For unconditional image synthesis on CIFAR-10 and CelebA, our modifications yield the best known FID scores on various GAN architectures. For conditional image synthesis on CIFAR-10, we improve the state-of-the-art FID score from 11.48 to 9.21. Finally, on ImageNet-2012, we apply our technique to the original BigGAN model and improve the FID from 6.66 to 5.38, which is the best score at that model size.

Downloads

Next from AAAI 2021

The Power of Literal Equivalence in Model Counting

Similar lecture

Fast and Scalable Adversarial Training of Kernel SVM via Doubly Stochastic Gradients

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from AAAI 2021

The Power of Literal Equivalence in Model Counting

Similar lecture

Fast and Scalable Adversarial Training of Kernel SVM via Doubly Stochastic Gradients

Downloads