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. Free, publicly-accessible full text available November 18, 2022
  2. ABSTRACT Extremely elongated, conducting dust particles (also known as metallic ‘needles’ or ‘whiskers’) are seen in carbonaceous chondrites and in samples brought back from the Itokawa asteroid. Their formation in protostellar nebulae and subsequent injection into the interstellar medium have been demonstrated, both experimentally and theoretically. Metallic needles have been suggested to explain a wide variety of astrophysical phenomena, ranging from the mid-infrared interstellar extinction at $\sim \,$3–8$\, {\rm \mu m}$ to the thermalization of starlight to generate the cosmic microwave background. To validate (or invalidate) these suggestions, an accurate knowledge of the optics (e.g. the amplitude and the wavelength dependence of the absorption cross sections) of metallic needles is crucial. Here we calculate the absorption cross sections of iron needles of various aspect ratios over a wide wavelength range, by exploiting the discrete dipole approximation, the most powerful technique for rigorously calculating the optics of irregular or nonspherical grains. Our calculations support the earlier findings that the antenna theory and the Rayleigh approximation, which are often taken to approximate the optical properties of metallic needles, are indeed inapplicable.
  3. 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 SP 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.
  4. 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 can 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.
  5. 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 can 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.
  6. Free, publicly-accessible full text available August 1, 2023
  7. 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 will 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.