Seminars

See all upcoming seminars in LäsIT and seminar web pages at the homepage for the PhD studentseminars, TDB, Vi2, Theory and Applications Seminars (TAS) @ UpMARC., Department of Mathematics and The Stockholm Logic Seminar.

Seminar
Today
Guido Tack, Monash University, Melbourne, Australia: An Abstract-Machine Model for MiniZinc
Location: ITC 1311, Time: 15:15-16:00

MiniZinc is a declarative modelling language for constrained optimisation. At its core, it is a first-order language that allows users to introduce existentially quantified decision variables (over types such as Booleans, integers, and floats) and express constraints using built-in predicates, functions, and Boolean connectives. On top of this, users can define their own predicates and functions. Evaluation of a MiniZinc model happens in two steps: first, the MiniZinc system performs "flattening" of the model, which produces an equivalent model that only consists of variable declarations and simple constraints.

In the second step, a solver takes the flat model and produces solutions (fixing decision variables to values that satisfy all constraints). Solvers are external tools that may use a variety of algorithmic techniques, such as Constraint Programming, Mixed Integer Programming, or Local Search. The "flattening" performed by the MiniZinc system essentially treats MiniZinc as a functional programming language, where the side effect of evaluating a function is to emit constraints that are added to the generated flat model. The current MiniZinc system implements this evaluation using an interpreter. In this talk, I will present our first steps towards replacing this interpreter with an intermediate representation and corresponding abstract machine model, which will achieve higher translation speed and at the same time produce better flat models.

Pedagogical lunch seminar
Tomorrow (15 Oct)
Matteo Magnani: Efficient teaching
Location: ITC 4308, Time: 12:00-13:00

The link for registration to get lunch (the same for all the seminars) is: https://doodle.com/poll/br8nbsgcvi7tycvy (if you do not want us to arrange lunch for you, you are welcome to come without registering).

Seminar
Thursday 17 Oct
Irina Tezaur, Sandia National Laboratories, USA: Albany Land-Ice (ALI): A Next-Generation Variable-Resolution Ice Sheet Model Towards Probabilistic Projections of Sea-Level Change
Location: ITC 2345, Time: 13:15

Guest seminar
Thursday 17 Oct
James Hensman, PROWLER.io: Variational inference in Gaussian processes; with application to forecasting in supply chain management
Location: Häggsalen, Ångströmlaboratoriet, Time: 14:15-15:00

Welcome to a guest lecture in the master course Advanced probabilistic machine learning by James Hensman. James is a senior machine learning scientist at PROWLER.io with more than fifty peer-reviewed publications. He leads the company’s research in the area of probabilistic models.

Disputation | Dissertation
Friday 18 Oct
Gong Cheng: Numerical ice sheet modeling: Forward and inverse problems
Location: ITC 2446, Time: 10:15

Gong Cheng will present and defend his PhD thesis Numerical ice sheet modeling: Forward and inverse problems.
Opponent: PhD Irina Tezaur,Sandia National Laboratories in Livermore, CA, USA.
Supervisors: Lina von Sydow, Per Lötstedt, Nina Kirchner,.

DiVA includes an abstract and the full text of the thesis.

Licentiatseminarium | Licentia
Friday 18 Oct
Carl Jidling: Tailoring Gaussian processes for tomographic reconstruction
Location: ITC 1211, Time: 13:15

A probabilistic model reasons about physical quantities as random variables that can be estimated from measured data. The Gaussian process is a respected member of this family, being a flexible non-parametric method that has proven strong capabilities in modelling a wide range of nonlinear functions. This thesis focuses on advanced Gaussian process techniques; the contribution consist of practical methodologies primarily intended for inverse tomographic applications.

In our most theoretical formulation, we propose a constructive procedure for building a customised covariance function given any set of linear constraints. These are explicitly incorporated in the prior distribution and thereby guaranteed to be fulfilled by the prediction.

One such construction is employed for strain field reconstruction, to which end we successfully introduce the Gaussian process framework. A particularly well-suited spectral based approximation method is used to obtain a significant reduction of the computational load. The formulation has seen several subsequent extensions, represented in this thesis by a generalisation that includes boundary information and uses variational inference to overcome the challenge provided by a nonlinear measurement model.

We also consider X-ray computed tomography, a field of high importance primarily due to its central role in medical treatments. We use the Gaussian process to provide an alternative interpretation of traditional algorithms and demonstrate promising experimental results. Moreover, we turn our focus to deep kernel learning, a special construction in which the expressiveness of a standard covariance function is increased through a neural network input transformation. We develop a method that makes this approach computationally feasible for integral measurements, and the results indicate a high potential for computed tomography problems.

Seminar
Friday 18 Oct
Luca Mottola: The What and What Not of Intermittent Computing
Location: ITC 1311, Time: 14:15-15:00

Abstract
Energy harvesting and wireless energy transfer are laying the foundations for a battery-less Internet of Things (IoT). These forms of energy provisioning are generally erratic across space and time. Executions become intermittent, as they consist of intervals of active computation interleaved by periods of recharging energy buffers. This trait challenges established practices at designing, implementing, and testing IoT systems, requiring a conceptual as well as practical leap in both hardware and software. Fundamental computing concepts such as consistency of data and progression of time need to be revisited. In this talk, I will elicit the key features of intermittent computing systems, discuss the current state of the art in the field, and outline open problems and long-term challenges still to be tackled.

IT20 Talk
23 October
Carolina Wählby, Stefan Engblom mfl.: Count on your brain - how can IT improve health?
Location: Siegbahnsalen, Ångström, Time: 13:15-14:15

Anmäl dig till seminariet

Pedagogical lunch seminar
12 November
Matteo Magnani: Assessment
Location: ITC 4308, Time: 12:00-13:00

The link for registration to get lunch (the same for all the seminars) is: https://doodle.com/poll/br8nbsgcvi7tycvy (if you do not want us to arrange lunch for you, you are welcome to come without registering).

Seminar
22 November
Luca Mottola: Mobile Drone Computing
Location: ITC 1311, Time: 14:15-15:00

Abstract
Autonomous drones are emerging as a powerful new breed of mobile sensing platform. Small embedded computers that move almost unconstrained while carrying rich sensor payloads, such as cameras and microphones, bring sensing where no other technology can reach. Notwithstanding recent advancements, the current use of drone technology is often limited to manual control of individual devices by skilled individuals. The gap between this and large-scale autonomous operation remains significant. The challenges span diverse disciplines, up to regulations and legal aspects. In this talk, we discuss the research efforts we are carrying out to overcome some of these challenges. Our work includes programming and operating systems, software verification, and flight control. We deployed many of these solutions in a range of real-world applications, including aerial photogrammetry and 3D reconstruction. We conclude by pointing out the several open problems in this field.

See also the list of all upcoming seminars.

Internal seminars. Lecturers may be either internal or external.

RSS