Page 214 - DCAP310_INTRODUCTION_TO_ARTIFICIAL_INTELLIGENCE_AND_EXPERT_SYSTEMS

Introduction to Artificial Intelligence & Expert Systems

Notes 11.3 Partial Match Techniques

Many different measures for evaluating the performance of information retrieval systems have
been proposed. The measures require a collection of documents and a query. All common
measures described here assume a ground truth notion of relevancy: every document is known
to be either relevant or non-relevant to a particular query. In practice, queries may be ill-posed
and there may be different shades of relevancy.
Precision is the fraction of the documents retrieved that are relevant to the user’s information
need. In binary classification, precision is analogous to positive predictive value. Precision
takes all retrieved documents into account. It can also be evaluated at a given cut-off rank,
considering only the topmost results returned by the system. Note that the meaning and usage
of “precision” in the field of Information Retrieval differs from the definition of accuracy and
precision within other branches of science and technology.
Recall is the fraction of the documents that are relevant to the query that are successfully retrieved.
In binary classification, recall is often called sensitivity. So it can be looked at as the probability
that a relevant document is retrieved by the query.

It is trivial to achieve recall of 100% by returning all documents in response to any query.
Therefore recall alone is not enough but one needs to measure the number of non-relevant
documents also, for example by computing the precision.
The proportion of non-relevant documents that are retrieved, out of all non-relevant documents
available.
11.3.1 Feature-based Techniques

The feature-based techniques extract local regions of interest (features) from the images and
identify the corresponding features in each image of the sequence. Such algorithms have initially
been developed for tracking a small number of salient features in long image sequences. The
tracking process can be divided into two major subtasks: feature extraction and feature tracking.
One can extract features only in the first image and search for the same features in the subsequent
images. This dynamic feature extraction scheme is advocated by the KLT Tracker. Alternatively,
one can find features in each static image prior to tracking and then find the corresponding
features along the sequence. Most of the feature tracking algorithms, including our IPAN Tracker,
operate in this way, although the dynamic feature extraction seems to be more natural and
reliable than the static one: the temporal information helps decide where and how to search for
features in the next frame.

In order to make use of features in processing geometric models, some core common aspects
exist. These apply irrespective of the domain of problems and the particular problems at hand.
The relevance of individual aspects may vary. These are as follows:
1. Identication of Features: A particular feature of interest in a given 3D model must be
specified in a way such that it can be algorithmically identified in the 3D data. This
definition can be a mix of topological properties, geometric metrics, colour and texture
attributes and so on. Some higher-level features may be defined in terms of simpler low-
level features. Such hierarchies of features are common in literature.
2. Recognition of Features: Recognition of a feature in the given 3D model is an essential
computational part feature-based processing of geometric models. For detecting an
individual model, typically specialised filters need to be implemented.
3. Suppressing Features: By suppressing a feature we mean removing a local instance of the
feature while minimally disturbing the data around the feature.

208 LOVELY PROFESSIONAL UNIVERSITY