<div dir="ltr">





<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)">Data Science</p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)">Academic Seminar</p>
<p class="gmail-p2" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0);min-height:14px"><br></p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)">March 24, 2022</p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)">12:00 p.m. de São Paulo, Brasil (UTC/GMT -03:00)</p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)"><a href="https://zoom.us/j/97410173464">https://zoom.us/j/97410173464</a><span class="gmail-Apple-converted-space"> </span></p>
<p class="gmail-p2" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0);min-height:14px"><br></p>
<p class="gmail-p2" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0);min-height:14px"><br></p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)">Speaker: Helio Migon</p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)">University: Universidade Federal do Rio de Janeiro - UFRJ</p>
<p class="gmail-p2" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0);min-height:14px"><br></p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)">Title: k-PARAMETRIC DYNAMIC GENERALIZED LINEAR MODELS</p>
<p class="gmail-p2" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0);min-height:14px"><br></p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)"><span class="gmail-Apple-converted-space"> </span>Abstract:<span class="gmail-Apple-converted-space"> </span></p>
<p class="gmail-p1" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0)"><span class="gmail-Apple-converted-space"> </span>Dynamic generalised linear models may be seen simultaneously as an extension to dynamic linear models and to generalised linear models, formally treating serial auto-correlation inherent to responses observed through time.<span class="gmail-Apple-converted-space">  </span>The current presentation revisits the Dynamic Linear Model class, focusing on specific concerns such as monitoring and intervention analysis. An approach based on information geometry is used to extend the DGLM to the k-parametric exponential family. The suggested technique supports multinomial responses and may be extended to accommodate compositional responses on k = d + 1 categories, while maintaining the sequential feature of the Bayesian inferential procedure, resulting in real-time inference. The updating scheme<span class="gmail-Apple-converted-space">  </span>takes advantage of the exponential family's conjugate structure, which ensures computing efficiency. The method is based on concepts like Kullback-Leibler divergence and the projection theorem, putting it in line with modern approaches to variational inference. We will show applications to real-world data to demonstrate the applicability of the proposed strategy. The method's computational efficiency is demonstrated in comparison to other approaches, as well as its flexibility to quickly accommodate new information when strategically needed, all while maintaining aspects of monitoring and intervention analysis, as well as discount factors, which are common in sequential analyses.</p>
<p class="gmail-p2" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0);min-height:14px"><br></p>
<p class="gmail-p2" style="margin:0px;font-variant-numeric:normal;font-variant-east-asian:normal;font-stretch:normal;font-size:12px;line-height:normal;font-family:Helvetica;color:rgb(0,0,0);min-height:14px"><span class="gmail-Apple-converted-space"> </span></p><div><br></div></div>