Nuno Paulino

Luís Miguel Sousa, Nuno Paulino, João Canas Ferreira et al.

Decision trees are often preferred when implementing Machine Learning in embedded systems for their simplicity and scalability. Hoeffding Trees are a type of Decision Trees that take advantage of the Hoeffding Bound to allow them to learn patterns in data without having to continuously store the data samples for future reprocessing. This makes them especially suitable for deployment on embedded devices. In this work we highlight the features of an HLS implementation of the Hoeffding Tree. The implementation parameters include the feature size of the samples (D), the number of output classes (K), and the maximum number of nodes to which the tree is allowed to grow (Nd). We target a Xilinx MPSoC ZCU102, and evaluate: the design's resource requirements and clock frequency for different numbers of classes and feature size, the execution time on several synthetic datasets of varying sample sizes (N), number of output classes and the execution time and accuracy for two datasets from UCI. For a problem size of D3, K5, and N40000, a single decision tree operating at 103MHz is capable of 8.3x faster inference than the 1.2GHz ARM Cortex-A53 core. Compared to a reference implementation of the Hoeffding tree, we achieve comparable classification accuracy for the UCI datasets.

Helder F. S. Campos, Nuno Paulino

This paper presents and discusses an implementation of a multiple target tracking method, which is able to deal with target interactions and prevent tracker failures due to hijacking. The referenced approach uses a Markov Chain Monte Carlo (MCMC) sampling step to evaluate the filter and constructs an efficient proposal density to generate new samples. This density integrates target interaction terms based on Markov Random Fields (MRFs) generated per time step. The MRFs model the interactions between targets in an attempt to reduce tracking ambiguity that typical particle filters suffer from when tracking multiple targets. A test sequence of 662 grayscale frames containing 20 interacting ants in a confined space was used to test both the proposed approach and a set of importance sampling based independent particle filters, to establish a performance comparison. It is shown that the implemented approach of modeling target interactions using MRF successfully corrects many of the tracking errors made by the independent, interaction unaware, particle filters.

Hélder Campos, Nuno Paulino

This paper presents results of a study of the performance of several base classifiers for recognition of handwritten characters of the modern Latin alphabet. Base classification performance is further enhanced by utilizing Viterbi error correction by determining the Viterbi sequence. Hidden Markov Models (HMMs) models exploit relationships between letters within a word to determine the most likely sequence of characters. Four base classifiers are studied along with eight feature sets extracted from the handwritten dataset. The best classification performance after correction was 89.8%, and the average was 68.1%