Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userSPMoE: Generate Multiple Pattern-Aware Outputs with Sparse Pattern Mixture of Experts
135
0
0.0
(
0
)
Added by
Shaobo Cui
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Many generation tasks follow a one-to-many mapping relationship: each input could be associated with multiple outputs. Existing methods like Conditional Variational AutoEncoder(CVAE) employ a latent variable to model this one-to-many relationship. However, this high-dimensional and dense latent variable lacks explainability and usually leads to poor and uncontrollable generations. In this paper, we innovatively introduce the linguistic concept of pattern to decompose the one-to-many mapping into multiple one-to-one mappings and further propose a model named Sparse Pattern Mixture of Experts(SPMoE). Each one-to-one mapping is associated with a conditional generation pattern and is modeled with an expert in SPMoE. To ensure each language pattern can be exclusively handled with an expert model for better explainability and diversity, a sparse mechanism is employed to coordinate all the expert models in SPMoE. We assess the performance of our SPMoE on the paraphrase generation task and the experiment results prove that SPMoE can achieve a good balance in terms of quality, pattern-level diversity, and corpus-level diversity.
rate research
Read More
Sparsely-gated Mixture of Experts networks (MoEs) have demonstrated excellent scalability in Natural Language Processing. In Computer Vision, however, almost all performant networks are dense, that is, every input is processed by every parameter. We present a Vision MoE (V-MoE), a sparse version of the Vision Transformer, that is scalable and competitive with the largest dense networks. When applied to image recognition, V-MoE matches the performance of state-of-the-art networks, while requiring as little as half of the compute at inference time. Further, we propose an extension to the routing algorithm that can prioritize subsets of each input across the entire batch, leading to adaptive per-image compute. This allows V-MoE to trade-off performance and compute smoothly at test-time. Finally, we demonstrate the potential of V-MoE to scale vision models, and train a 15B parameter model that attains 90.35% on ImageNet.
Domain generalizable (DG) person re-identification (ReID) is a challenging problem because we cannot access any unseen target domain data during training. Almost all the existing DG ReID methods follow the same pipeline where they use a hybrid dataset from multiple source domains for training, and then directly apply the trained model to the unseen target domains for testing. These methods often neglect individual source domains discriminative characteristics and their relevances w.r.t. the unseen target domains, though both of which can be leveraged to help the models generalization. To handle the above two issues, we propose a novel method called the relevance-aware mixture of experts (RaMoE), using an effective voting-based mixture mechanism to dynamically leverage source domains diverse characteristics to improve the models generalization. Specifically, we propose a decorrelation loss to make the source domain networks (experts) keep the diversity and discriminability of individual domains characteristics. Besides, we design a voting network to adaptively integrate all the experts features into the more generalizable aggregated features with domain relevance. Considering the target domains invisibility during training, we propose a novel learning-to-learn algorithm combined with our relation alignment loss to update the voting network. Extensive experiments demonstrate that our proposed RaMoE outperforms the state-of-the-art methods.
Query rewriting (QR) systems are widely used to reduce the friction caused by errors in a spoken language understanding pipeline. However, the underlying supervised models require a large number of labeled pairs, and these pairs are hard and costly to be collected. Therefore, We propose an augmentation framework that learns patterns from existing training pairs and generates rewrite candidates from rewrite labels inversely to compensate for insufficient QR training data. The proposed framework casts the augmentation problem as a sequence-to-sequence generation task and enforces the optimization process with a policy gradient technique for controllable rewarding. This approach goes beyond the traditional heuristics or rule-based augmentation methods and is not constrained to generate predefined patterns of swapping/replacing words. Our experimental results show its effectiveness compared with a fully trained QR baseline and demonstrate its potential application in boosting the QR performance on low-resource domains or locales.
We investigate the problem of deterministic pattern matching in multiple streams. In this model, one symbol arrives at a time and is associated with one of s streaming texts. The task at each time step is to report if there is a new match between a fixed pattern of length m and a newly updated stream. As is usual in the streaming context, the goal is to use as little space as possible while still reporting matches quickly. We give almost matching upper and lower space bounds for three distinct pattern matching problems. For exact matching we show that the problem can be solved in constant time per arriving symbol and O(m+s) words of space. For the k-mismatch and k-difference problems we give O(k) time solutions that require O(m+ks) words of space. In all three cases we also give space lower bounds which show our methods are optimal up to a single logarithmic factor. Finally we set out a number of open problems related to this new model for pattern matching.
Simultaneous machine translation (SiMT) generates translation before reading the entire source sentence and hence it has to trade off between translation quality and latency. To fulfill the requirements of different translation quality and latency in practical applications, the previous methods usually need to train multiple SiMT models for different latency levels, resulting in large computational costs. In this paper, we propose a universal SiMT model with Mixture-of-Experts Wait-k Policy to achieve the best translation quality under arbitrary latency with only one trained model. Specifically, our method employs multi-head attention to accomplish the mixture of experts where each head is treated as a wait-k expert with its own waiting words number, and given a test latency and source inputs, the weights of the experts are accordingly adjusted to produce the best translation. Experiments on three datasets show that our method outperforms all the strong baselines under different latency, including the state-of-the-art adaptive policy.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
