What’s the difference between cognitive computing and predictive analytics?
Predictive analytics uses numerical data or statistical distributions to populate a mathematical model or algorithm, and through varying the inputs to the model, an output is derived. Assuming the numerical data used has a predictive quality i.e. temperature variation within one standard deviation, then the model will output a numeric answer within probability or accuracy. Optimization and stochastic techniques can be applied to the model to account for variations in the data and adjust the output toward a weighted variable.
Cognitive computing utilizes three core components of Artificial Intelligence (AI) that form the foundational differences between predictive analytics. These components are understanding, reasoning and learning. Decades of research and applied technology has gone into the formulation and improvement of patented algorithms and processing techniques that mimic human thought and reasoning. These capabilities are being actively applied to augment human decision-making in a variety of industries and applications. Unlike predictive analytics that depends largely on structured data, cognitive computing uses the written word or unstructured data as its primary source for insight. Cognitive computing solutions leverage these AI components and natural language processing to review and weigh all the relevant textual evidence within source documents. By understanding context and rapidly processing the corpus of unstructured data against defined rules, questions can be asked and answers can be rendered with scored probability of accuracy based on the weight of the evidence provided. In summary, cognitive computing can understand, reason and learn through natural language processing, it can make inferences, detect sediment and understand context then provide evidence-based answers that humans can use to make their final decisions.
How did IBM correlate cognitive computing in healthcare to applications in aircraft diagnostics?
IBM’s organization and culture is set up to correlate technologies and solutions across industries. In this case, our A&D team recognized the parallels in using unstructured data associated with diagnosing a patient’s condition and applying these same cognitive processes to a physical asset like an aircraft. The aerospace industry, similar to healthcare, is focused on determining the prognostic health of their respective patients (aircraft, systems and sub-systems) primary through the acquisition of numeric data produced by the sensors, instruments and controls onboard. A similar challenge exists between the two industries where much of the information about the patient’s history, health, maintenance and repair, check-ups, relevant research, alerts, logs etc. are captured in textual form but not typically correlated with the numeric information. Cognitive computing provides the capability to bring this pool of ‘dark’ textual, unstructured data in to the light to leverage it’s for the benefit of the patient’s overall health.
Do you have trial programs in place? If so, with whom do they involve and how are they coming along?
We have numerous examples across several industry sectors where the cognitive capabilities described above are being applied in trial and production programs. Several of these cognitive programs have been publically disclosed such as Memorial Sloan Kettering Institute for support of oncology diagnosis, HR Block to ensure every tax deduction and credit is found, Airbus to customize aircraft maintenance. The adoption of cognitive capabilities is a digital transformation journey. IBM is building a scalable cognitive platform requiring a sound cloud-based infrastructure, open source tools and APIs within a development environment to support a dynamically growing ecosystem of non-IBM and IBM developers.
What challenges do you foresee that may hinder or stall adoption of cognitive computing in aviation maintenance?
Cognitive is a disruptive technology and we suggest that any company considering cognitive start by asking themselves these four questions:
* Is there proof that with its adoption, the cost of operations will be reduced and margins will be improved by an order of magnitude that is compelling?
* What is the level of effort required to fully adopt the technology?
* Is there a well thought-out platform and ecosystem approach making the adoption relatively easily, and is it supported by a company(s) that enables the technology to scale?
* What do we risk by not adopting the technology?
IBM has numerous use cases across global industries where clients are seeing compelling improvement to operating costs and margins when cognitive computing as applied. Taking a crawl, walk, run approach to adoption is the best way to build a viable roadmap for success. Finally, we are seeing adoption in the aviation maintenance domain and other peer industries today. We believe it’s only a matter of time before the rest of the industry sees it as well.
What new, future technologies are you most excited about and why?
One of our recent acquisitions was The Weather Company (TWC), which is not a future technology but one that IBM is very excited about. TWC represents the largest IoT data platforms on the planet. Supporting 127K global sensing stations, 40M mobile phones, and 50K flights/day with real-time weather, atmospheric, air quality, global lighting, air traffic and incident data. This new Internet of Things/cognitive platform represents a strategic, digital technology available now to support the big data challenge over the next few decades the Aerospace Industry must meet.
In the category of new and exciting technologies we see a lot of potential with Blockchain. In simple terms, Blockchain is a distributed, shared ledger technology allowing any authorized participant(s) in the secure business network or community to see customized or complete views of transactions and processes where there are large and complex interdependencies. A good example of where we see this gaining acceptance and adoption is in the realm of transportation and logistics. We believe that this capability has great promise within the complex aerospace supply chains process and offers substantial reductions in transaction times, overhead costs of intermediaries, and creates end-to-end transparency of all constituents within a secure community network.