b-bit Marginal Regression
Authors: Martin Slawski, Ping Li
NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We here provide numerical results supporting/illustrating some of the key points made in the previous sections. We also compare b-bit Marginal Regression to alternative recovery algorithms. Setup. Our simulations follow model (1) with n = 500, s {10, 20, . . ., 50}, σ {0, 1, 2} and b {1, 2}. ... The experiments reveal that what is predicted by the analysis of the comparison of the relative performance of 1-bit and 2-bit measurements for estimating x closely agrees with what is observed empirically, as can be seen in Figure 2. |
| Researcher Affiliation | Academia | Martin Slawski Department of Statistics and Biostatistics Department of Computer Science Rutgers University martin.slawski@rutgers.edu Ping Li Department of Statistics and Biostatistics Department of Computer Science Rutgers University pingli@stat.rutgers.edu |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described. It mentions a "Supplement" for proofs and additional experiments, but no specific statement about code availability. |
| Open Datasets | No | The paper does not provide concrete access information for a publicly available or open dataset. Instead, it describes generating synthetic data for its simulations: "Our simulations follow model (1) with n = 500, s {10, 20, . . ., 50}, σ {0, 1, 2} and b {1, 2}. Regarding x , the support and its signs are selected uniformly at random, while the absolute magnitude of the entries corresponding to the support are drawn from the uniform distribution on [β, 2β]..." |
| Dataset Splits | No | The paper mentions that "Each possible configuration for s, f and σ is replicated 20 times." but does not specify train, validation, or test dataset splits (e.g., percentages, sample counts, or cross-validation setup). |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library or solver names with version numbers, needed to replicate the experiment. |
| Experiment Setup | Yes | Our simulations follow model (1) with n = 500, s {10, 20, . . ., 50}, σ {0, 1, 2} and b {1, 2}. Regarding x , the support and its signs are selected uniformly at random, while the absolute magnitude of the entries corresponding to the support are drawn from the uniform distribution on [β, 2β], where β = f (1/λ1,σ) p log(n)/m and m = f 2(1/λ1,σ)2s log n with f {1.5, 3, 4.5, . . ., 12} controlling the signal strength. The resulting signal is then normalized to unit 2-norm. ... For b = 2, we use Lloyd-Max quantization for a N(0, 1)-random variable which is optimal for σ = 0, but not for σ > 0. Each possible configuration for s, f and σ is replicated 20 times. |