One surprising trait of neural networks is the extent to which their connections can be pruned with little to no effect on accuracy. But when we cross a critical level of parameter sparsity, pruning any further leads to a sudden drop in accuracy. This drop plausibly reflects a loss in model complexity, which we aim to avoid. In this work, we explore how sparsity also affects the geometry of the linear regions defined by a neural network, and consequently reduces the expected maximum number of linear regions based on the architecture. We observe that pruning affects accuracy similarly to how sparsity affects the number of linear regions and our proposed bound for the maximum number. Conversely, we find out that selecting the sparsity across layers to maximize our bound very often improves accuracy in comparison to pruning as much with the same sparsity in all layers, thereby providing us guidance on where to prune.

No abstract

Medicinal plants are used as traditional treatment for numerous human diseases. According to WHO, medicinal plants would be the best source to obtain a variety of drugs. Contrary to the synthetic drugs, antimicrobial of plant origin are not associated with many side effects and have an enormous therapeutic potential to heal many infectious diseases. Present study was carried out on the screening of some medicinal plants against selected pathogenic organisms in the period of October 2017 to January 2018. A total of nine different medicinal plants were screened and evaluated for their antimicrobial activity against 10 bacterial species. Among them, Euphorbia hirta, Azadirachtaindica, Artemisia vulgaris were found to be effective against gram positive bacteria (Bacillus subtilis, Staphylococcus aureus, Methicillin resistant Staphylococcus aureus MRSA), whereas rest six medicinal plants (Aeglemarmelos, Justiciaadhatoda, Ficusreligiosa, Syzygiumcumini, Nyctanthes arbor and Meliaazedarach) were found to be ineffective against all the microorganisms (Bacillus subtilis, Escherichia coli, Klebsiellapneumoniae, Proteus vulgaris, Pseudomonas aerogenosa, Salmonella typhii, Shigelladysentriae, MRSA). The minimum bactericidal concentration (MBC) of Euphorbia hirta against Bacillus subtilis and Staphylococcus aureus was found to be 12.5mg/ml and MBC of Artemisia vulgaris against Bacillus subtilis and MRSA was also found to be 12.5 mg/ml while of Staphylococcus aureus was 25mg/ml. Similarly, the MBC of Azadirachta indica against Staphylococcus aureus was 25mg/ml.