-
The Lack of Convexity of the Relevance-Compression Function
Authors:
Albert E Parker,
Tomas Gedeon,
Alexander Dimitrov
Abstract:
In this paper we investigate the convexity of the relevance-compression function for the Information Bottleneck and the Information Distortion problems. This curve is an analog of the rate-distortion curve, which is convex. In the problems we discuss in this paper, the distortion function is not a linear function of the quantizer, and the relevance-compression function is not necessarily convex (c…
▽ More
In this paper we investigate the convexity of the relevance-compression function for the Information Bottleneck and the Information Distortion problems. This curve is an analog of the rate-distortion curve, which is convex. In the problems we discuss in this paper, the distortion function is not a linear function of the quantizer, and the relevance-compression function is not necessarily convex (concave), but can change its convexity. We relate this phenomena with existence of first order phase transitions in the corresponding Lagrangian as a function of the annealing parameter.
△ Less
Submitted 22 April, 2022;
originally announced April 2022.
-
Mapping and Validating a Point Neuron Model on Intel's Neuromorphic Hardware Loihi
Authors:
Srijanie Dey,
Alexander Dimitrov
Abstract:
Neuromorphic hardware is based on emulating the natural biological structure of the brain. Since its computational model is similar to standard neural models, it could serve as a computational acceleration for research projects in the field of neuroscience and artificial intelligence, including biomedical applications. However, in order to exploit this new generation of computer chips, rigorous si…
▽ More
Neuromorphic hardware is based on emulating the natural biological structure of the brain. Since its computational model is similar to standard neural models, it could serve as a computational acceleration for research projects in the field of neuroscience and artificial intelligence, including biomedical applications. However, in order to exploit this new generation of computer chips, rigorous simulation and consequent validation of brain-based experimental data is imperative. In this work, we investigate the potential of Intel's fifth generation neuromorphic chip - `Loihi', which is based on the novel idea of Spiking Neural Networks (SNNs) emulating the neurons in the brain. The work is implemented in context of simulating the Leaky Integrate and Fire (LIF) models based on the mouse primary visual cortex matched to a rich data set of anatomical, physiological and behavioral constraints. Simulations on the classical hardware serve as the validation platform for the neuromorphic implementation. We find that Loihi replicates classical simulations very efficiently and scales notably well in terms of both time and energy performance as the networks get larger.
△ Less
Submitted 22 September, 2021;
originally announced September 2021.
-
FlowME: Lattice-based Traffic Measurement
Authors:
Petko Valtchev,
Omar Mounaouar,
Omar Cherkaoui,
Alexandar Dimitrov,
Laurent Marchand
Abstract:
Flow-based traffic measurement is a very challenging problem: Managing counters for each individual traffic flow in hardware resources knowingly struggle to scale with high-speed links. In this paper we propose a novel lattice theory-based approach that improves flow-based measurement performances and scales by keeping the number of the maintained hardware counters to a minimum (result mathematica…
▽ More
Flow-based traffic measurement is a very challenging problem: Managing counters for each individual traffic flow in hardware resources knowingly struggle to scale with high-speed links. In this paper we propose a novel lattice theory-based approach that improves flow-based measurement performances and scales by keeping the number of the maintained hardware counters to a minimum (result mathematically established in the paper). The crucial contribution of the lattice is to map the computational semantics of the packet processing to user requests for traffic measurement thus allowing for a better-informed and focused counter assignment.
An implementation over an Openflow switch, FlowME, was developed and evaluated upon its memory usage, performance overhead, and processing effort to generate the minimal solution. Experimental results indicate a significant decrease in resource consumption.
△ Less
Submitted 11 November, 2012;
originally announced November 2012.
