Næringslivsnettverket KiD inviterer til fagkveld med temaet maskinlæring og big data. Kom for å høre hvordan bedrifter bruker disse teknologiene, delta på konkurranse med kule premier og spis god mat i hyggelig selskap.

Hva?

Denne gangen er det SINTEF Digital som er hovedarrangør for kvelden, og de får med seg et knippe andre bedrifter fra nettverket. Flere av disse vil holde foredrag som gir deg innsikt i hva maskinlæring og big data er og hvordan det brukes av IT-bedrifter i Norge. Etter foredrag blir det case-konkurranse og servering av mat.

Mer om programmet:

SINTEF

Title: Predicting fish-catch in space and time: eSushi

SINTEF, in cooperation with Nordnes AS, Dualog AS, Furuno Norge AS og Havfisk AS has used historical data from fishing vessels to create a model of where to catch fish. We will quickly go though some experiences, results and show a demo of the system.

Telenor

Title: Unsupervised Learning: Make Sense of your (not labelled) data

​Abstract: In many real scenario application, we need to handle data that are not labelled. As a first step, it is useful to extract patterns and make sense of the data we are going to use and manipulate. This talk will present what are the most recent advancement in this area (and in particular in the context of Deep Learning), including some concrete examples from different domains.

EVRY

Title: Do you want AI that answers customer service questions? You can train one in 15 minutes.

Abstract: Audience can find out how we in EVRY used the data from our ‘Customer service desk’ to train the algorithms which are able to understand the user’s problem, provide the appropriate answer and/or implement the corresponding solution.

The presentation would include live coding demonstration that is supposed to encourage and inspire the students to further explore the topic.

The content will include some (most, depending on time) of the following:

• Basics of natural language processing (simple, yet interesting approaches used in text processing, e.g. tokenization, stemming, vectorization of text)
• Topic classification using machine(deep) learning. I would quickly train and compare the performance of different algorithms on a small dataset.
• Usage of trained algorithms to provide answers to users’ questions
• Optional (if there is time) Semantic models (word2vec and/or Glove). How pre-trained word-vectors can help us understand semantic relationships.

Computas

Title: Predicting future bus delays with TensorFlow in the cloud

Abstract: We have recently completed a bigdata machine learning project based on the technology stack from Google. We would like to present and discuss our experience. The task was to build a forecast system that predicts future bus delays for the Rogaland public bus service company Kolumbus. We would like to show our data pipelines for learning and prediction and which tools and APIs from Google Cloud Plattform and Python proved to be useful. You will get an insight and first-hand experience on how to program TensorFlow with Keras and how to ingest and rig in place a large datasets Ajusing Apache Beam. Spoiler: Data Science projects are 80% data cleansing, and 10% machine learning.

Etter foredragene vil det være bespisning og en case-konkurranse der Skagerak Kraft og Avinor bidrar med hver sin oppgave. Dere vil få veiledning av bedriftsrepresentantene underveis og det vil være premier.