skip to main content

Search for: All records

Creators/Authors contains: "Liu, F."

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. We propose a new Scaled Population (SP) based arithmetic computation approach that achieves considerable improvements over existing stochastic computing (SC) techniques. First, SP arithmetic introduces scaling operations that significantly reduce the numerical errors as compared to SC. Experiments show accuracy improvements of a single multiplication and addition operation by 6.3X and 4X, respectively. Secondly, SP arithmetic erases the inherent serialization associated with stochastic computing, thereby significantly improves the computational delays. We design each of the operations of SP arithmetic to take O(1) gate delays, and eliminate the need of serially iterating over the bits of the population vector. Our SPmore »approach improves the area, delay and power compared with conventional stochastic computing on an FPGA-based implementation. We also apply our SP scheme on a handwritten digit recognition application (MNIST), improving the recognition accuracy by 32.79% compared to SC.« less
  2. Brain imaging genetics is an important research topic in brain science, which combines genetic variations and brain structures or functions to uncover the genetic basis of brain disorders. Imaging data collected by different technologies, measuring the same brain distinctly, might carry complementary but different information. Unfortunately, we do not know the extent to which phenotypic variance is shared among multiple imaging modalities, which might trace back to the complex genetic mechanism. In this study, we propose a novel dirty multi-task SCCA to analyze imaging genetics problems with multiple modalities of brain imaging quantitative traits (QTs) involved. The proposed method canmore »not only identify the shared SNPs and QTs across multiple modalities, but also identify the modality-specific SNPs and QTs, showing a flexible capability of discovering the complex multi-SNP-multi-QT associations. Compared with the multi-view SCCA and multi-task SCCA, our method shows better canonical correlation coefficients and canonical weights on both synthetic and real neuroimaging genetic data. This demonstrates that the proposed dirty multi-task SCCA could be a meaningful and powerful alternative method in multi-modal brain imaging genetics.« less
  3. This paper proposes a new meta-learning method – named HARMLESS (HAwkes Relational Meta LEarning method for Short Sequences) for learning heterogeneous point process models from short event sequence data along with a relational network. Specifically, we propose a hierarchical Bayesian mixture Hawkes process model, which naturally incorporates the relational information among sequences into point process modeling. Compared with existing methods, our model can capture the underlying mixed-community patterns of the relational network, which simultaneously encourages knowledge sharing among sequences and facilitates adaptive learning for each individual sequence. We further propose an efficient stochastic variational meta expectation maximization algorithm that canmore »scale to large problems. Numerical experiments on both synthetic and real data show that HARMLESS outperforms existing methods in terms of predicting the future events.« less
  4. Model compression is significant for the wide adoption of Recurrent Neural Networks (RNNs) in both user devices possessing limited resources and business clusters requiring quick responses to large-scale service requests. This work aims to learn structurally-sparse Long Short-Term Memory (LSTM) by reducing the sizes of basic structures within LSTM units, including input updates, gates, hidden states, cell states and outputs. Independently reducing the sizes of basic structures can result in inconsistent dimensions among them, and consequently, end up with invalid LSTM units. To overcome the problem, we propose Intrinsic Sparse Structures (ISS) in LSTMs. Removing a component of ISS willmore »simultaneously decrease the sizes of all basic structures by one and thereby always maintain the dimension consistency. By learning ISS within LSTM units, the obtained LSTMs remain regular while having much smaller basic structures. Based on group Lasso regularization, our method achieves 10:59 speedup without losing any perplexity of a language modeling of Penn TreeBank dataset. It is also successfully evaluated through a compact model with only 2:69M weights for machine Question Answering of SQuAD dataset. Our approach is successfully extended to non-LSTM RNNs, like Recurrent Highway Networks (RHNs). Our source code is available.« less
  5. Topic extraction presents challenges for the bibliometric community, and its performance still depends on human intervention and its practical areas. This paper proposes a novel kernel k-means clustering method incorporated with a word embedding model to create a solution that effectively extracts topics from bibliometric data. The experimental results ofa comparison of this method with four clustering baselines (i.e., k-means, fuzzy c-means, principal component analysis, and topic models) on two bibliometric datasets demonstrate its effectiveness across either a relatively broad range of disciplines or a given domain. An empirical study on bibliometric topic extraction from articles published by three top-tiermore »bibliometric journals between 2000 and 2017, supported by expert knowledge-based evaluations, provides supplemental evidence of the method’s ability on topic extraction. Additionally, this empirical analysis reveals insights into both overlapping and diverse research interests among the three journals that would benefit journal publishers, editorial boards, and research communities.« less
  6. Free, publicly-accessible full text available March 1, 2023
  7. Free, publicly-accessible full text available December 1, 2022