BME Seminar Series will take place on Tuesday, October 23rd at 12:30 pm - 1:20 pm in MSB 384.

**Yue Zhou** will discuss “Development of a Wearable Tremor Suppression Glove", **Parsa Omidi** will present his work on “Algorithm for phase-displacement conversion from digital holograms", and **Jacob Tryon** will present “Evaluating Performance of EEG/EMG Fusion Methods for Motion Classification”.

**Speaker:** Lila Kari (Waterloo)

**Title:** Machine Learning and the Mathematics of Genomes

**Abstract:**

In the same way we use the twenty-six letters of the alphabet to write text, and the two bits 0 and 1 to write computer code, the four basic DNA units (Adenine, Cytosine, Guanine, Thymine) are used by Nature to encode information as DNA strands. Theoretically, a DNA strand can be viewed as a "word” over the four-letter alphabet {A, C, G, T}, and the mathematical structure of such words has implications for their biological structure and function.

This talk describes our research into the mathematical properties of genomic DNA sequences by exploring the connection between word frequencies in a genome and the type of organism that the genome belongs to. In particular, I describe our investigation into the Chaos Game Representation of a DNA sequence as a potential "genomic signature” of its species. Moreover, I describe how we combine supervised machine learning techniques with such genomic signatures for ultrafast, accurate, and scalable algorithms for species identication and classication. The potential impact of such alignment-free universal classication algorithms could be signicant, given that 86% of existing species on Earth and 91% of species in the oceans still await classication.

All are welcome. Coffee and cookies will be served.

]]>**Speaker:** Mario Ghossoub

**Title:** Optimal Insurance Design: Belief Heterogeneity and Ambiguity

**Abstract:**

We re-examine the problem of demand for insurance indemnification when the insured and the insurer disagree about the likelihoods associated with the realizations of the insurable loss. For ease of comparison with the classical literature, we adopt the original setting of Arrow, but allow for divergence in beliefs between the insurer and the insured. We do not impose the no sabotage condition on admissible indemnities. Instead, we impose a state-verification cost that the insurer can incur in order to verify the loss severity, which rules out ex post moral hazard issues that could otherwise arise from possible misreporting of the loss by the insured. Moreover, unlike the existing literature, we do not impose conditions on the type or level of disagreement about probabilities, and we allow in particular for singularity between the beliefs, that is, disagreement about zero-probability events. We characterize the optimal indemnity for any type or level of belief heterogeneity, and we show that it has a simple two-part structure: full insurance on an event to which the insurer assigns zero probability, and a variable deductible on the complement of this event. We then examine the effect of ambiguity in beliefs on the shape of optimal indemnities, and we provide a closed-form characterization optimal indemnities in the case of unilateral and bilateral probability weighting, thereby extending several results in the literature.

**Speaker:** Christoph Frei (Alberta)

**Speaker:** Kaisa Miettinen (U of Jyvaskyla, Finland)