Home > When Neurons Fail - Technical Report > HTML MARC |

000217561 001__ 217561 000217561 005__ 20190118220112.0 000217561 037__ $$aREP_WORK 000217561 245__ $$aWhen Neurons Fail - Technical Report 000217561 260__ $$bEPFL$$c2016 000217561 269__ $$a2016 000217561 300__ $$a19 000217561 336__ $$aWorking Papers 000217561 500__ $$aPresented at the 2016 Biological Distributed Algorithms workshop (http://www.snl.salk.edu/~navlakha/BDA2016/) To Appear in IPDPS 2017 as a regular paper. 000217561 520__ $$aNeural networks have been traditionally considered robust in the sense that their precision degrades gracefully with the failure of neurons and can be compensated by additional learning phases. Nevertheless, critical applications for which neural networks are now appealing solutions, cannot afford any additional learning at run-time. In this paper, we view a multilayer neural network as a distributed system of which neurons can fail independently, and we evaluate its robustness in the absence of any (recovery) learning phase. We give tight bounds on the number of neurons that can fail without harming the result of a computation. To determine our bounds, we leverage the fact that neural activation functions are Lipschitz-continuous. Our bound is on a quantity, we call the Forward Error Propagation, capturing how much error is propagated by a neural network when a given number of components is failing, computing this quantity only requires looking at the topology of the network, while experimentally assessing the robustness of a network requires the costly experiment of looking at all the possible inputs and testing all the possible configurations of the network corresponding to different failure situations, facing a discouraging combinatorial explosion. We distinguish the case of neurons that can fail and stop their activity (crashed neurons) from the case of neurons that can fail by transmitting arbitrary values (Byzantine neurons). In the crash case, our bound involves the number of neurons per layer, the Lipschitz constant of the neural activation function, the number of failing neurons, the synaptic weights and the depth of the layer where the failure occurred. In the case of Byzantine failures, our bound involves, in addition, the synaptic transmission capacity. Interestingly, as we show in the paper, our bound can easily be extended to the case where synapses can fail. We present three applications of our results. The first is a quantification of the effect of memory cost reduction on the accuracy of a neural network. The second is a quantification of the amount of information any neuron needs from its preceding layer, enabling thereby a boosting scheme that prevents neurons from waiting for unnecessary signals. Our third application is a quantification of the trade-off between neural networks robustness and learning cost. 000217561 6531_ $$aneural networks 000217561 6531_ $$adistributed computing 000217561 6531_ $$afault tolerance 000217561 6531_ $$aapproximate computing 000217561 6531_ $$amachine learning 000217561 6531_ $$abiological distributed algorithms 000217561 700__ $$0246705$$aEl Mhamdi, El Mahdi$$g200613 000217561 700__ $$0240335$$aGuerraoui, Rachid$$g105326 000217561 8560_ $$felmahdi.elmhamdi@epfl.ch 000217561 8564_ $$s521221$$uhttps://infoscience.epfl.ch/record/217561/files/When_Neurons_Fail_1.pdf 000217561 8564_ $$s1048098$$uhttps://infoscience.epfl.ch/record/217561/files/When_Neurons_Fail_Tech%20Report.pdf$$yn/a$$zn/a 000217561 8564_ $$s551531$$uhttps://infoscience.epfl.ch/record/217561/files/When_Neurons_Fail___Third_Attempt.pdf$$yn/a$$zn/a 000217561 8564_ $$s566801$$uhttps://infoscience.epfl.ch/record/217561/files/When_Neurons_Technical_Report.pdf$$yn/a$$zn/a 000217561 909C0 $$0252114$$pDCL$$xU10407 000217561 909CO $$ooai:infoscience.tind.io:217561$$pIC$$qworking 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 917Z8 $$x200613 000217561 937__ $$aEPFL-WORKING-217561 000217561 973__ $$aEPFL$$rREVIEWED$$sPUBLISHED 000217561 980__ $$aREP_WORK