Knowledge mining sensory evaluation data: genetic programming, statistical techniques, and swarm optimization

Knowledge mining sensory evaluation data is a challenging process due to extreme sparsity of the data, and a large variation in responses from different members (called assessors) of the panel. The main goals of knowledge mining in sensory sciences are understanding the dependency of the perceived liking score on the concentration levels of flavors’ ingredients, identifying ingredients that drive liking, segmenting the panel into groups with similar liking preferences and optimizing flavors to maximize liking per group. Our approach employs (1) Genetic programming (symbolic regression) and ensemble methods to generate multiple diverse explanations of assessor liking preferences with confidence information; (2) statistical techniques to extrapolate using the produced ensembles to unobserved regions of the flavor space, and segment the assessors into groups which either have the same propensity to like flavors, or are driven by the same ingredients; and (3) two-objective swarm optimization to identify flavors which are well and consistently liked by a selected segment of assessors.     

Bayesian hypothesis testing for pattern discrimination in brain decoding

Research in cognitive neuroscience and in brain–computer interfaces (BCI) is frequently concerned with finding evidence that a given brain area processes, or encodes, given stimuli. Experiments based on neuroimaging techniques consist of a stimulation protocol presented to a subject while his or her brain activity is being recorded. The question is then whether there is enough evidence of brain activity related to the stimuli within the recorded data. Finding a link between brain activity and stimuli has recently been proposed as a classification task, called brain decoding. A classifier that can accurately predict which stimuli were presented to the subject provides support for a positive answer to the question. However, it is only the answer for a given data set and the question still remains whether it is a general rule that will apply also to new data. In this paper we try to reliably answer the neuroscientific question about the presence of a significant link between brain activity and stimuli once we have the classification results. The proposed method is based on a Beta-Binomial model for the population of generalization errors of classifiers from multi-subject studies within the Bayesian hypothesis testing framework. We present an application on nine brain decoding investigations from a real functional magnetic resonance imaging (fMRI) experiment about the relation between mental calculation and eye movements.     

Adaptive classifiers for multisource OCR

When patterns occur in large groups generated by a single source (style consistent test data), the statistics of the test data differ from those of the training data, which consist of patterns from all sources. We present a Gaussian model for continuously distributed sources under which we develop adaptive classifiers that specialize in the statistics of style-consistent test data. On NIST handwritten digit data, the adaptive classifiers reduce the error rate by more than 50% operating on one writer ( samples/class) at a time.Received: 14 November 2002, Accepted: 6 March 2003, Published online: 12 September 2003Correspondence to: George Nagy     

3D-printed and foamed triply periodic minimal surface lattice structures for energy absorption applications

Structures currently used for energy absorption include foams, composites, and honeycombs. Recent studies have indicated the potential of triply periodic minimal surfaces (TPMS) for energy absorption applications as well as weight reduction. This study presents three TPMS lattice structures, namely the Gyroid, Fischer-Koch S, and PMY, which are fabricated in uniform and graded densities. These structures, created in the MSLattice software are 3D printed using polylactic acid. Subsequently, the 3D-printed structures undergo a gas foaming process to investigate benefits of higher porosity on energy absorption. The structures are then characterized for porosity, compressive properties, energy absorption, and thermal properties. The results show that the uniform and graded density structures have similar energy absorption values as long as the structures have a similar average density with the PMY structure exhibiting the highest energy absorption value, both in unfoamed and foamed conditions, respectively. The foaming process increased the porosity by 50% but did not improve the energy absorption characteristics of any of the structures with the foamed PMY structures exhibiting the least deviation compared to the unfoamed samples. These foamed TPMS structures are suitable for applications in the automotive and aerospace industry that demand lightweight structures for energy absorption.     

Efficient implementation of mixed-precision multiply-accumulator unit for AI algorithms

Many of the modern deep learning, machine learning, and artificial intelligence algorithms use adders, multipliers, and multiply-accumulators (MACs) with mixed precisions. In general, fixed-point and floating-point adders, multipliers, and MACs are used in mixed-precision hardware, such that the above algorithm can choose appropriate hardware for its processing. This paper proposes an efficient mixed-precision MAC circuit that can perform fixed-point and floating-point operations. The proposed design uses Han-Carlson adder with late carry or end-around carry in its accumulator design; as a result, delay and energy of the circuit reduce by and respectively, when compared with the existing designs from the literature.     

An adaptive multimodal biometric management algorithm

This paper presents an evolutionary approach to the sensor management of a biometric security system that improves robustness. Multiple biometrics are fused at the decision level to support a system that can meet more challenging and varying accuracy requirements as well as address user needs such as ease of use and universality better than a single biometric system or static multimodal biometric system. The decision fusion rules are adapted to meet the varying system needs by particle swarm optimization, which is an evolutionary algorithm. This paper focuses on the details of this new sensor management algorithm and demonstrates its effectiveness. The evolutionary nature of adaptive, multimodal biometric management (AMBM) allows it to react in pseudoreal time to changing security needs as well as user needs. Error weights are modified to reflect the security and user needs of the system. The AMBM algorithm selects the fusion rule and sensor operating points to optimize system performance in terms of accuracy.     

A 5–7 GHz current reuse and g<Subscript>m</Subscript>-boosted common gate low noise amplifier with LC based ESD protection in 32 nm CMOS

Scaling of minimum length of the MOSFET has improved its performance but has reduced the breakdown voltage which makes it prone to Electrostatic Discharge (ESD) damage. This work presents a low-power g _m -boosted common gate (CG) ultra wideband (UWB) low noise amplifier (LNA) architecture, operating in the 5–7 GHz range, employing current-reuse technique with LC based Electrostatic Discharge (ESD) protection. Common gate topology supports wide band input matching and noise figure independent of operating frequency. A PMOS common source topology is used as the g_m-boosting stage in order to reduce the noise figure and to remove the dependency of noise figure from the bias point. The g_m-boosting stage and the amplifier share common bias current to reduce the power consumption of the LNA. A shunt inductor, series capacitor and power clamp are used for protecting the circuit from ESD damage. The ESD circuit is co-designed with the input matching network in order to reduce the area of the layout. The proposed topology has shown significant improvement in gain and noise figure with ESD protection.