Many industrial optimisation problems involve the challenging task of efficiently searching for optimal decisions from a huge set of possible combinations. The optimal solution is the one that best optimises a set of objectives or goals, such as maximising productivity while minimising costs. If we have a nice mathematical equation for how each objective depends on the decisions we make, then we can usually employ standard mathematical approaches, such as calculus, to find the optimal solution. But what do we do when we have no idea how our decisions affect the objectives, and thus no equations? What if all we have is a small set of experiments, where we have tried to measure the effect of some decisions? How do we make use of this limited information to try to find the best decisions?

This talk will present a common industrial optimisation problem, known as expensive black box optimisation, through a case study from the manufacturing sector. For problems like this, calculus can’t help, and trial and error is not an option! We will introduce some methods and tools for tackling expensive black-box optimisation. Finally, we will discuss new methodologies for assessing the strengths and weaknesses of optimisation methods, to ensure the right method is selected for the right problem.

This will be a whistle-stop tour of a few topics on infectious disease modelling, mainly influenza. Topics to include:

• challenges in capturing dynamics of pathogens with multiple co-circulating strains
• untangling the 2009 influenza pandemic from medical insurance claims data from the US
• bioinformatic methods to detect viral packaging signals
• and a big science project (top secret until the talk!)

Julia will be visiting the Mathematical Institute on sabbatical this term, and hopes this talk will help us find areas of overlapping interests.

Image use continues to increase in both biomedical sciences and clinical practice. State of the art acquisition techniques allow characterisation from subcellular to whole organ scale, providing quantitative information of structure and function. In the heart, for example, images acquired from a single modality (cardiac MRI) can characterise micro- and macrostructure, describe mechanical function and measure blood flow. In the lungs, new contrast agents can be used to visualise the flow of gas in free breathing subjects. This provides rich new sources of information as well as new challenges to extract data in a way that is useful to clinicians as well as computer modellers.
I will describe efforts in my group to use the latest advances in machine learning to analyse images, and explain how we are applying these to the development of accurate computer models of the heart.
 

Feedback control is found extensively in many natural and technological systems. Indeed, many biological processes use feedback
to regulate key processes – examples include bacterial chemotaxis and negative autoregulation in genetic circuits. Despite the prevalence of
feedback in natural systems, its design and implementation in a Synthetic Biological context is much harder.  In this talk I will give
examples of how we implemented feedback systems in three different biological systems. The first one concerns the design of a synthetic
recombinase-based feedback loop, which results into robust expression. The second describes the use of small RNAs to post-transcriptionally
regulate gene expression through interaction with messenger RNA (mRNA). The third involves the introduction of negative feedback in a
two-component signalling system through a controllable phosphatase.  Closing, I will outline the challenges posed by the design of such
systems, both theoretical and on their implementation.

Current medical practice is driven by the experience of clinicians, by the difficulties of integrating enormous amounts of complex and heterogeneous static and dynamic data and by clinical guidelines designed for the “average” patient. In this talk, I will describe some of my research on developing novel, specially-crafted machine learning theories, methods and systems aimed at extracting actionable intelligence from the wide variety of information that is becoming available (in electronic health records and elsewhere) and enabling every aspect of medical care to be personalized to the patient at hand. Because of the unique and complex characteristics of medical data and medical questions, many familiar machine-learning methods are inadequate.  My work therefore develops and applies novel machine learning theory and methods to construct risk scores, early warning systems and clinical decision support systems for screening and diagnosis and for prognosis and treatment.  This work achieves enormous improvements over current clinical practice and over existing state-of-the-art machine learning methods.  By design, these systems are easily interpretable and so allow clinicians to extract from data the necessary knowledge and representations to derive data-driven medical epistemology and to permit easy adoption in hospitals and clinical practice. My team has collaborated with researchers and clinicians in oncology, emergency care, cardiology, transplantation, internal medicine, etc. You can find more information about our past research in this area at: http://medianetlab.ee.ucla.edu/MedAdvance.