D. Charlebois

Gene expression noise facilitates adaptation and drug resistance independently of mutation.

We show that the effect of stress on the reproductive fitness of noisy cell populations can be modeled as a first-passage time problem, and demonstrate that even relatively short-lived fluctuations in gene expression can ensure the long-term survival of a drug-resistant population. We examine how this effect contributes to the development of drug-resistant cancer cells, and demonstrate that permanent immunity can arise independently of mutations.

Automating reuse of software for expert system analysis of remote sensing data

Systems involving remote sensing analysis for airborne and satellite data in combination with geographic information systems are large and complex. The Canada Centre for Remote Sensing (CCRS) has created an expert system shell and several expert systems in order to provide image analysis programs with the necessary knowledge to solve difficult image processing problems, such as updating a forest inventory geographic information system. An interactive task interface (ILTI) provides an expert system with a Prolog module designed to answer queries from the image analysis program by retrieving knowledge from an image analysis knowledge base, the analyst advisor. Image analysis experts currently create ILTIs. They have found this to be a time-consuming task. An incremental/adaptive planner has been developed that will create a plan that emulates the ILTIs behavior by analyzing image processing session dialogues between a human expert and an image analysis program for several cases of forest updates. The planner relies on a knowledge base in order to generalize and modify plans acquired from session dialogues. The planner speeds and simplifies the creation of new expert systems. >

Machine learning from remote sensing analysis

An intelligent system (SEIDAM-System of Experts for Intelligent Data Management) is being developed for answering queries about the forests and the environment through the integration of remote sensing, geographic information, models and field measurements. SEIDAM consists of an hierarchical group of expert systems. Machine learning and planning can be used to create plans that execute image analysis software in order to recognize specific objects and perform a variety of different tasks. A query (task) could require, for example, that forest inventory stored in a GIS be updated to reflect past harvesting. As sensors become more numerous, the choices of data and options to recognize objects become more complex. These complexities can be reduced by making use of case-based reasoning. Only the data needed to answer the query will be used. The aim of case-based reasoning is to avoid having to build a solution to a problem from first principles, or by drawing on rare expertise, by adapting a known solution for an old problem to the new problem. There is a constantly growing variety of data sets. Providing information at various degrees of accuracy and at often different cost. A non-expert user of this data could greatly benefit from reusing specific cases of queries, which convert the data into knowledge that other users were seeking before. A case, in this context, consists of a query and an example of the process (plan) that answers that query using a single or a multi-sensor data set, and geographic information, such as forest cover, topography, hydrology, etc. In their earlier work, the authors constructed a planner (LEAR), a planning system for creating expert systems by executing software for a case and interacting with a human expert. they now wish to raise the machine learning methods from creating expert systems for executing existing software to creating new rules (knowledge) derived from observing remote sensing analysis cases. Knowledge about objects acquired by the LEAR planner can be used to assist a case-based reasoner during both its retrieval step and its adaptation step.<<ETX>>

Automated forest inventory update with SEIDAM

As part of the Applied Information Systems Research Program sponsored by NASA, a System of Experts for Intelligent Data Management, SEIDAM, has been created. As a component of SEIDAM, a case-based reasoning system called PALERMO was developed in order to reason about the process of digital forest inventory update. SEIDAM uses a set of software agents that carry out tasks such as translate point elevation data into a digital terrain model or import polygonal information from a geographical information file into an image format or ingest remote sensing data and update meta data databases. In this paper the authors discuss the new agents that were created for automatic classification and the ease with which they were added to the SEIDAM environment.

Case-based reasoning and software agents for intelligent forest information management

To perform forest information management, SEIDAM integrates forest cover descriptions, topographic maps and remote sensing imagery. SEIDAM relies on an online robotic data storage device, image and GIS metadata databases, software agents and a case-based reasoning system to deliver information to decision makers in a timely fashion. The image and GIS metadata databases contain information about the sources of data, where the data are stored, where they have been delivered and the processing they have undergone. The software agents perform the actual processing by running image analysis, GIS, database and other software to accomplish specific tasks. The case-based reasoning system relies on the software agents, past experience from domain experts and information from the metadata databases to determine what processing is required to deliver products satisfying user goals. This paper describes the intelligent inventory update function in SEIDAM and its AI methodology.

Machine learning and planning for data management in forestry

The Seidam project uses an AI planning-based approach that combines three problem-solving methods-transformational analogy, derivational analogy and goal regression-to automatically answer forest-management queries. The project is conducted under NASAs Applied Information Systems Research Program. Seidam, which runs on a Sun Sparcstation using the Solaris 2.3 version of Unix, is a complex system that relies on extensive cooperation between expert systems and processing agents.

An Algorithm for the Stochastic Simulation of Gene Expression and Heterogeneous Population Dynamics

We present an algorithm for the stochastic simulation of gene expression and heterogeneous population dynamics. The algorithm combines an exact method to simulate molecular-level fluctuations in single cells and a constant-number Monte Carlo method to simulate time-dependent statistical characteristics of growing cell populations. To benchmark performance, we compare simulation results with steady-state and time-dependent analytical solutions for several scenarios, including steady-state and time-dependent gene expression, and the effects on population heterogeneity of cell growth, division, and DNA replication. This comparison demonstrates that the algorithm provides an efficient and accurate approach to simulate how complex biological features influence gene expression. We also use the algorithm to model gene expression dynamics within “bet-hedging” cell populations during their adaption to environmental stress. These simulations indicate that the algorithm provides a framework suitable for simulating and analyzing realistic models of heterogeneous population dynamics combining molecular-level stochastic reaction kinetics, relevant physiological details and phenotypic variability.

Training agents in a complex environment

The paper describes an approach to building agents for users of complex data access and management systems for resource and environmental applications. Gathering good examples of this highly specialized and complicated activity is costly and difficult. There is usually only a small set of such good examples available to guide the development of an agent. Consequently, agents are trained, rather than being learned inductively from example sets. In our approach, agents use planning and plan generalization (learning) as their basic mechanism. Plans for yet unseen combinations of goals are created by the merging of plans for individual goals, with the minimum of replanning. An example illustrates merging of existing plans, and shows a simple practical solution to the mutual goal clobbering problem. Plans are built from low-granularity agent commands. The prototype of the system is implemented, and the paper shows a fragment of agent training. The application for this reasoning system addresses the use of planning and of agents to perform forest cover map updates using satellite imagery. To perform this task, a variety of geographical information systems, remote sensing image analysis tools and visualization packages are used.<<ETX>>

Case-based reasoning in an intelligent information system for forestry

Our objective is to integrate transformational analogy, derivational analogy, and goal- regression to create solutions for an intelligent system called SEIDAM (System of Experts for Intelligent Data Management). SEIDAM answers queries about forests and the environment through the integration of remote sensing, geographic information, models, and field measurements. A query (problem) could require, for example, that a forest inventory stored in a geographical information system be updated to reflect past harvesting by overlaying current satellite imagery over forest cover maps. A case consists of a query, remote sensing data, and geographic information, and the analysis methods to answer the query. SEIDAM will consist of approximately 150 expert systems performing satellite and aircraft image analysis, integrated to multiple GIS and a relational database. Derivational analogy provides the means by which this search can be expanded knowledgeably; i.e., provide a knowledge-based approach justifying the expansion of the search. Transformational analogy eliminates the problems associated with searching by foregoing a search altogether. The advantage is that the intractability of exploring the search space is no longer a consideration.

Planning and Learning in a Natural Resource Information System

The paper presents PALERMO — a planner used to answer queries in the SEIDAM information system for forestry. The information system is characterized by the large complexity of software and data sets involved. PALERMO uses previously answered queries and several planning techniques to put together plans that, when executed, produce products by calling the appropriate systems (GIS, image analysis, database, models) and ensures the proper flow on information between them. Experimental investigation of several planning techniques indicates that analogical planning cuts down the search involved in planning without experiencing the utility problem.