© Ellis Interstellar LLC 2017 (All Rights Reserved)
We have participated in several projects for idea generation contests sponsored by TopCoder and Innocentive. These include:
1. Exoplanet Data Taxomomy
Research, engineering techniques, and data analytics identify Goldilocks zones as potential destinations. The taxonomy is based on available data, anticipated growth in the number of known exoplanets, and a flexible engineering methodology. The foundation of the taxonomy is a classic trade study to rank and score candidate attributes based on criteria derived from a set of exoplanet taxonomy requirements. A "critic", a sequential optimizer, and software-based tests find a minimum spanning taxonomy. Software assures that, to the extent practical, each object has a unique path through the taxonomy.
Why a taxonomy? It is a valuable teaching tool, it organizes and sorts a data set, and a taxonomy is structured as a hierarchical characteristics-based ontology. This knowledge structure is readily captured in a knowledge base for multi-strategy reasoning and data analytics. That is the true value of a taxonomy! The taxonomy defines the parameter space of exoplanet characteristics. While some work "inside the box", others work outside the box", a taxonomy defines the box. We produced an upper-level exoplanet taxonomy. This taxonomy doesn't span the orbital, physical, environmental, and stellar characteristics of interest. Nor does it provide a unique set of characteristics for each of ~ 2,000 exoplanets currently identified. The Goldilocks Taxonomy has 13 parameters chosen from a collection of potential characteristics and provides fine-grained organization of exoplanet characteristics, a robust methodology to refine the taxonomy, and a suite of data analytics tools to discover interesting patterns.
2. Space Object Identification Taxomomy
The proposed taxonomy is based on Space Situation Awareness (SSA) insights from many years of work in the Space Control,
satellite owner & operator, and intelligence domains. These insights are the foundation for a classic trade study to rank
and score candidate nodes based on criteria derived from SSA missions and Challenge requirements. The nodes are further
processed using a “critic”, a sequential optimizer and rule induction to find a minimum spanning taxonomy.
At first glance, a taxonomy is an important way of visualizing that widely varying set of attributes
associated with a space object. As such, it is a valuable teaching tool that provides people interested in space with
an easy-to-understand view. Digging deeper, it provides a "world-view" of the domain that helps people understand it
in a more organized and holistic way. More practically, it provides a way to organize and sort on a data set, since
all the attributes important to a taxonomy are not, today, available in a single data file. Building on the data set
idea, a taxonomy is a hierarchical network and each node in the hierarchy has characteristics. This is the makings of
a truly powerful knowledge structure, or ontology. The knowledge structure is readily captured in a knowledge base such
as Protege. The resulting hierarchical, characteristics-based ontology supports multi-strategy reasoning and learning,
including simulation and planning! For further information, see Space Object Taxonomy.
3. Extent of the Placebo Effect
The challenge was to recruit a control group for clinical trials to quantify the extent of the placebo effect.
Each member of the control group makes a valuable contribution to the study, but represents a cost and a lost
opportunity to receive the treatment under evaluation to determine how the efficiency of generating statistically
relevant control group data can be enhanced. The objective is to systematically plan, administer, record and evaluate
(PARE) the number of placebos given in a clinical trial. For further information, see Placebo Effects.
4. Longevity Predictor
The challenge was to define a novel approach for predicting individual longevity. Provide sufficient detail for an
evaluator to understand the innovative aspects of the idea and appreciate its feasibility. For further information,
see Longevity Predictor.
5. New Approaches to Health Monitoring
Ingestable and epidermal sensors are paired with reasoning algorithms to manage health issues. New sensors provide enabling
technology, but the reasoning algorithms described here constitute the truly innovative aspect of the proposed idea. The
proposed solution is holistic because it addresses the wide variety of chemical, physical and psychological factors
associated with these health issues. Evidence is multi-modal because it is collected from ingested, epidermal, and
subvocalized or queried patient and caregiver responses for probabilistic combination. This paper describes three types
of sensors: