Escolar Documentos
Profissional Documentos
Cultura Documentos
Role of AI in Logistics
Presented at
Asst Prof. Deepankar Sinha AILABS Academy,
Indian Institute of Foreign Trade Kolkata on May 11th 2018
Use and distribution limited solely to authorized personnel. (c) Copyright 2018
• Tesla Car Crash
The key question
• Is this AI or Automation
Automation
PREDICT
PREDICT MAKE
MAKE
SOURCE
SOURCE DATA
DATA DECISION
DECISION
&&INFER
INFER
TAKE
TAKE
UPDATE
UPDATE DECISION
DECISION
AI
PREDICT MAKE
SOURCE DATA DECISION
& INFER
TAKE
UPDATE DECISION
Self Learning
Supervised Algorithms
• In supervised algorithms, we may not know the inner
relations of the data we are processing, but we do
know very well which is the output that we need from
our model. For example:
• “I need to be able to start predicting when users will
cancel their trip”.
• Notice that the output of we model is already defined:
“will vehicle X cancel its trip”. What we may not know
yet, is HOW to realise which vehicles will cancel. So we
can use an existing set of data to “train” a model into
predicting this particular aspect about our vehicles.
The training of the model
Supervised Algorithms - Example
• So for example, if we have 10,000 user’s usage history. From these,
maybe 5,000 cancelled and 5,000 have continued with the trip. So
what we can do, is take data from 4,500 users who cancelled, and
4,500 from vehicles which continued (data from 9000 users total).
• Train our model with this data, letting it “see” which cancelled and
which are using it.
• The model will do it’s best to predict the status of the user, and we
can compare with the real value. If out of the 1,000 users, the model
correctly predicted 891, then the model has a 89,1% accuracy.
Unsupervised Algorithms
• With unsupervised algorithms, we still don’t know what we want to
get out of the model yet. we probably suspect that there hast to be
some kinds of relationships or correlation between the data we have,
but data is too complex to try to guess.
• So in this cases we normalize our data into a format that makes sense
to compare, and then let the model work it’s magic and try to find
some of these relationships.