About Minerva AI
The rise of AI is likely to be one of the most significant trends in the technology sector over the coming years. Advances in AI are impacting companies of all sizes and in various sectors as businesses look to improve decision-making, reduce operating costs and enhance consumer experience. The concept of what defines AI has changed over time, but at its core is machines being able to perform tasks that would typically require human perception or cognition.
Recent breakthroughs in AI have been achieved by applying machine learning to very large data sets. However, machine learning – in particular deep learning systems – has limitations in that machine learning often fails when there is limited training data available or when the actual dataset differs from the training set. Also, it is often difficult to get clear explanations of the results produced by deep learning systems.
Minerva’s AI Platform addresses these limitations by using human generated knowledge models and probabilistic reasoning to generate predictive analyses. Because our system uses semantic networks as the primary data format we are able to provide explanations in natural language for our predictions as well as expert advice about what your next steps should be.
As the world continues to move to a state of interconnectivity, it behooves every organization – both public and private – to make sure their information is standardized so that it can be machine readable. Only once this state has been achieved will the world’s trove of historical data be truly accessible.
Relevance to Minerva
Artificial intelligence (AI) applications in the geosciences are constrained by interoperability problems. The use by researchers of different complex non-standard earth science taxonomies constitutes one of these problems. Different ways in which these taxonomies can be combined to build problem-domain ontologies, necessary for AI reasoning, constitutes another.
As data standardization is still in its infancy, Minerva is a strong supporter of and contributor to the development and maintenance of internationally curated vocabulary standards, such as the INSPIRE initiative in the European Union. Data interoperability standards for earth sciences established by INSPIRE address the problem of non-standard taxonomies. Minerva is actively working with INSPIRE to assist in creating data standards and is contributing to its improvement by identifying problems in the standards when the data is of insufficient quality for use with Minerva’s technology.
To learn more about earth science standardized taxonomies, please visit Commission for the Management and Application of Geoscience Information.
Minerva’s AI platforms rely on our proprietary software to build semantic networkA semantic network is a graph structure for representing knowledge in patterns of interconnected data points. graphs for clients’ data. These semantic structures can organize and condense information about concepts or objects which might otherwise be scattered in various databases or buried deep in the pages of a scientific paper.
Traditionally, the success of AI applications has often depended upon the data used:
- Is there enough?
- Is it appropriate?
- Is it of sufficient quality?
By using defined standards, semantic knowledge graphs help make data interpretable for humans and machines alike. Machines and algorithms make use of the semantic graphs to retrieve not only the objects themselves but also undiscovered relationships that can be found between the objects, even if they are not explicitly stated. Minerva’s AI Platforms allow ‘reasoning’ based on the embedded expert knowledge contained within the semantic networks.
For example, in Mineral Exploration, Minerva’s TERRA Mining AI employs semantic networks to describe mineral deposits, deposit models or other exploration targets. Semantic networks are also used to describe locations on the ground, or underground, in terms of all available information, then presents the data in a way that is clear, understandable, and auditable for users.
One of the most revolutionary components of Minerva’s AI technology is the powerful logical structure it applies to the knowledge with which it reasons. This is possible due to the ontological controlThe ability to define not only the vocabulary of data, but also the terminological, assertional, and relational axioms to define concepts, individuals, and roles employed by Minerva during knowledge capture. To enable meaningful results, a consistent vocabulary must be used for both the data and the predictions. This requires a domain-specific ontology so that terms are used consistently.
Minerva’s software creates taxonomiesA system for naming, defining, and classifying groups on the basis of shared characteristics. for each user’s dataset. Without such taxonomies, it would be almost impossible to engineer AI applications which emulate intelligent reasoning as effectively as a human expert might.
Geologists, for example, need to use scientific vocabulary to describe their exploration targets and the environments they occur in. The words in these vocabularies occur within sometimes complex taxonomies, such as the taxonomy of rocks, the taxonomy of minerals, and the taxonomy of geological time, to mention only a few. Minerva’s TERRA Mining AI Suite incorporates these taxonomies into their reasoning (they know, for example, that basalt is a volcanic rock, but granite is not).
For this reason, Minerva is a strong supporter of and contributor to the development and maintenance of internationally-curated vocabulary standards, such as the INSPIRE initiative in the European Union. Minerva is working with a number of the INSPIRE committees responsible for creating the standards for the data and is contributing to its improvement by identifying problems in the standards when the data is of insufficient quality for use with Minerva’s technology.
To learn more about earth science taxonomies, please visit Commission for the Management and Application of Geoscience Information.
Reasoning with Uncertainty
Just like people, artificial intelligence systems are inevitably forced to make decisions based on incomplete information. A doctor, for instance, cannot know exactly what is going on inside a patient without exploratory surgery, just as an exploration geologist cannot know exactly where best to look for, or expand the size of, a mineral deposit without physical exploration.
Minerva applies the principles of “reasoning with uncertainty” to the challenge of providing recommendations with the highest likelihood of success. In doing so, Minerva’s focus is on machine cognitionA neural network with the ability to take decisions systems, while incorporating outputs from machine perceptionThe capability of a computer system to interpret data in a manner that is similar to the way humans use their senses to relate to the world around them. algorithms, such as neural networks, for inclusion in its reasoning, as and when appropriate.