Politecnico di Milano
SchedaIncarico
 
Funzioni disponibili
Uscita

Risorse bibliografiche
Risorsa bibliografica obbligatoria
Risorsa bibliografica facoltativa
Scheda Riassuntiva
Anno Accademico 2015/2016
Scuola Scuola di Ingegneria Industriale e dell'Informazione
Insegnamento 097548 - NEUROENGINEERING [C.I.] - BIOE 475
Docente Cerveri Pietro , Pedrocchi Alessandra Laura Giulia
Cfu 10.00 Tipo insegnamento Corso Integrato


Corso di Studi Codice Piano di Studio preventivamente approvato Da (compreso) A (escluso) Insegnamento
Ing Ind - Inf (Mag.)(ord. 270) - MI (471) BIOMEDICAL ENGINEERING - INGEGNERIA BIOMEDICA*AZZZZ097548 - NEUROENGINEERING [C.I.] - BIOE 475

Programma dettagliato e risultati di apprendimento previsti

NEUROENGINEERING I

Aims of the class

This class aims at providing specific physiological insights about some high-level abilities of the human such as the reasoning, the pattern classification and the information storage. Different computational models, even mediated by physiologic knowledge, are described and analyzed to discover the potentiality and the drawbacks in the engineering representation of such abilities. Bayesian networks represent a mathematical tool that can be interpreted and adapted to the representation of reasoning ability. Neural networks, mathematical tools derived from the understanding of the single-neuron and population-neuron physiology, are shown to express different capabilities ranging from pattern classification to memorization of information.

Specific content

 Data classification by feed-forward neural networks. Main high-level physiological features of the neurons: plasticity and learning. McCulloc and Pitts formal neuron as a basic input/output information processing unit. Artificial neural networks as a mathematical tool for sub-symbolic computation. Network topology and architectures. Multi-layer perceptron demonstration of general ability to classify input patterns. Supervised methods to training neural networks: delta rule and back-propagation for feed-forward networks.

 Data storage by feed-back neural networks. Physiological memory physiology and models. Sensorial, short-term and long-term memory stores. Representing the episodic memory as a feed-back neural network: ability to store and retrieve knowledge. Dynamic properties of a feed-back neural network: convergence and stability. Auto-associative and hetero-associative memories. Hebbian learning (reinforcement learning) to estimate the neuron weights. Stability analysis of binary networks to memorize information patterns.

 Symbolic and sub-symbolic computation ability in the human brain. Introduction to high-level computational ability of the human mind from an information processing point of view. Human reasoning by manipulating and operating on symbols. Difference between symbolic and sub-symbolic operations by approaching the link to the physiological architecture of the brain.

 Bayesian approach to model uncertain reasoning. Uncertain reasoning: moving from probabilistic to causal representations. Limitations of the Bayesian statistics in modeling multiple causes and multiple effects. Bayesian networks and causal reasoning. Link between bayesian network topology and semantics by d-separation. Introducing the temporal dimension in the Bayesian networks: Hidden Markov models as a computational tool for systems evolving through time. Cellular automata as representation system including spatial and temporal dimensions.

Class organization

Besides, the presentation of the contents of the course in frontal lectures, the teaching will include frontal-based and computer-based exercices.

 

NEUROENGINEERING II

Aims of the class

The goal of the course is the study of neuroengineering methods and technologies, including algorithms such as unsupervised neuronal networks, models of computational motor control, design of neuroprostheses and robots for neurorehabilitation, and optical and electronics technologies for the reading and stimulation of in-vitro neuronal cultures.

Specific content

Artificial neural networks Unsupervised artificial neural networks. Self organizing maps.

Models for sensorimotor integration and neural basis of motor control. Motor learning, adaptation and re-learning after brain damage. Models of motor control system and sensorimotor integration derived from psycophysics, movement biomechanics and computational neurosciences. Neural basis for motor control in humans. Motor re-learning after brain damage: neurorehabilitation issues.

Neurorobotics and Functional Electrical Neuromuscular Stimulation Robotics in rehabilitation: introduction and control strategies. Functional Electrical Stimulation in neurorehabilitation: bioengineering issues. Biomimetic control systems for neuroprostheses. Interfacing and control, utilizing artificial neural networks for Functional Electrical Stimulation. Applications in rehabilitation.

In vitro Neuroengineering Basic in neuroengineering. Neuron on chip: perspectives in biotechnology. Spike detection: algorithms and experimental studies.

Class organization

Besides, the presentation of the contents of the course in frontal lectures, the teaching will include practical lessons, practical activities in laboratories, visits to external laboratories and clinical premises and a discussion teaching workshop.

This latter is organized into small groups (about 8 students), led by a tutor (usually a PhD student or a Post-doc). Each group deeply studies a given research topic, starting from a couple of research papers proposed by the tutor, discusses it and defines a goal of a project to improve what presented in the literature. In a second session, the group will discuss a feasibility study of a solution to the given problem. The final solution, shared by the group, is presented to the all class into a final all-together workshop.

The participation to the Discussion Teaching is optional, those who participate won’t have to take the oral part of the Neuroengineering 2 exam (except for oral with both professors for high marks).

 

 

 


Bibliografia

Mix Forme Didattiche
Tipo Forma Didattica Ore didattiche
lezione
33.0
esercitazione
10.0
laboratorio informatico
9.0
laboratorio sperimentale
0.0
progetto
0.0
laboratorio di progetto
0.0

Informazioni in lingua inglese a supporto dell'internazionalizzazione
Insegnamento erogato in lingua Inglese
DisponibilitÓ di materiale didattico/slides in lingua inglese
PossibilitÓ di sostenere l'esame in lingua inglese
DisponibilitÓ di supporto didattico in lingua inglese

Note Sulla ModalitÓ di valutazione

The student will have to take two written tests (one for each part of the exam) and an oral exam. In the case the averaged mark of the two written parts will be equal or above 27/30 the student will take the oral with both professors, spreading over the both parts. Otherwise, two separate orals are required

28/04/2017 Area Servizi ICT v. 1.2.1 / 1.2.1