IBM Deep Learning Service

Artificial Intelligence as a Service

“…Apart from the fact that users should have comprehensive knowledge and in-depth experience for an optimal setup and configuration, the process itself is challenging and time-consuming (Duong and Sang 2018). Users need to manage physical and virtual machines and install required AI libraries, which is more challenging in the context of AI because users have to ensure the resilience of the training jobs and facilitate consistent response times for inference requests, among others (Bhattacharjee et al 2017). Consequently, AIaaS spares users considerable complexity as they bypass setup and configuration and transfer this task (and related risks) to the AIaaS provider.…”

“…AI software and developer services also support users through automatic hyper-parameter tuning, hence further optimizing the performance of the AI algorithm. Besides popular automatic tuning approaches, such as Random search and Bayesian optimization (Wang et al 2018), accessing observations of performance and characteristics of previously trained models allows providers to improve the automatic tuning of hyper-parameters even more (Bhattacharjee et al 2017). Hereby, providers analyze historical data (across their users) to understand which hyper-parameter configurations yielded satisfactory results in the past.…”

Medical Machines: The Expanding Role of Ethics in Technology-Driven Healthcare

“…A recent clinical trial organized in the United States demonstrated an AI system's ability to detect diabetic retinopathy with specialty-level skill from ophthalmological images of real patients' eyes (25). Watson, a supercomputer developed by IBM Corporation, uses semantic technology and deep learning to find applications in (among other places) the healthcare system, applied across spheres of research, diagnosis, and therapeutic decision making (26)(27)(28). Watson for Oncology, a collaboration between Memorial Sloan Kettering Cancer Centre and IBM, applies machine learning to existing data sets in order to determine how it can best recommend treatment for a variety of cancers; recently, this technology has demonstrated remarkable concordance with professional tumour boards (29).…”

A Comprehensive Tutorial and Survey of Applications of Deep Learning for Cyber Security

“…DL as a service is provided by various cloud platform such as Microsoft Azure ML [149], Google Cloud ML Platform [150], Amazon ML [151], IBM Watson Analytics [152] and so on. IBM's DLaaS software architecture details are provided in [153]. Virtual machines are one of the important components in infrastructure as a service (IaaS).…”