From Bayesian Sparsity to Gated Recurrent Nets
Authors: Hao He, Bo Xin, Satoshi Ikehata, David Wipf
NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The resulting insights lead to a novel sparse estimation system that, when granted training data, can estimate optimal solutions efficiently in regimes where other algorithms fail, including practical direction-of-arrival (DOA) and 3D geometry recovery problems. [...] This section presents experiments involving synthetic data and two applications. [...] Figures 3(a) and 3(b) evaluate our model, averaged across 105 trials, against an array of optimization-based approaches: SBL [33], ℓ1 norm minimization [4], and IHT [5]; and existing learning-based DNN models: an ISTA-inspired network [20], an IHT-inspired network [34], and the best maximal sparsity net (Max Sparse Net) from [38] (detailed settings in the supplementary). |
| Researcher Affiliation | Collaboration | Hao He Massachusetts Institute of Technology haohe@mit.edu; Bo Xin Microsoft Research, Beijing, China jimxinbo@gmail.com; Satoshi Ikehata National Institute of Informatics satoshi.ikehata@gmail.com; David Wipf Microsoft Research, Beijing, China davidwipf@gmail.com |
| Pseudocode | No | The paper describes algorithms and derivations using mathematical equations and prose but does not include any explicitly labeled pseudocode blocks or algorithms. |
| Open Source Code | No | The paper does not contain any explicit statements or links indicating that open-source code for the described methodology is provided or will be made available. |
| Open Datasets | No | The paper describes a 'novel online data generation process' for synthetic data: 'to create samples in each mini-batch, we first generate a vector x with random support pattern and nonzero amplitudes. We then compute y = Φx + ϵ, where ϵ is a small Gaussian noise component.' For application data, it refers to literature for how data is computed or generated (e.g., 'see supplementary for background and details on how to compute Φ'), but does not provide direct links to publicly available datasets or access to their generated data. |
| Dataset Splits | No | The paper does not explicitly provide details about training/validation/test dataset splits, specific percentages, or absolute sample counts for these splits. It mentions training using an online data generation process and evaluating on generated or specified problem instances. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments, such as GPU or CPU models, or cloud computing specifications. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers, such as programming languages, libraries, or frameworks used for implementation. |
| Experiment Setup | No | The paper mentions that 'a separate network must be trained via SGD' and that 'Full network details are deferred to the supplementary', but it does not provide specific hyperparameter values (e.g., learning rate, batch size, number of epochs) or detailed training configurations in the main text. |