Machine translators are regularly used to translate text from one language to another. In spite of the advances that have been made in this field, the output may still need to be proofread by a human.

Describe two problems that these translators may encounter when translating from one language to another.

An old set of 2D animated cartoons from the 1940s has been discovered and it is decided to modify them to turn them into 3D animation.

Global warming is a term used to describe the increase in mean global temperatures.

There have been numerous computer simulations developed to predict the effects of global warming. One simulation is NASA’s Virtual Earth System Laboratory (VESL), which allows users to see how climate change affects glacier size, global sea level and changes to the coastline.

The VESL runs simulations in real time and is an abstraction of reality.

A recent study reported that previous simulations of predicted global sea levels for 2100 were highly inaccurate.

Weather forecasters use computer models which are able to simulate future weather patterns. These forecasts were originally limited to the near future. However, modern systems can now produce long range forecasts.

The simulation of the weather forecasting models produces specific data which can be output in a variety of ways.

Google Translate is an algorithm whose function is to translate text from one language to another. One of the resources that it uses is the body of documents produced by the United Nations which are routinely translated by humans into various languages.

Discuss the reasons why Google Translate takes a probabilistic approach in preference to a cognitive rule-based approach.

The electronic control unit (ECU) of a vehicle is an on-board computer that is constantly monitoring the performance of several components of a vehicle. For example, the ECU:

• controls the functioning of the lights, the brakes, the airbag, and the fuel-level signal
• permits the scale being changed in some digital displays, such as switching the speedometer from miles/hour to km/hour.

The software embedded in an ECU receives input data from a variety of sources when it runs auto-diagnostic tests.

Neural networks are often used to model the complex behaviour of systems. These networks must first be trained in order to establish a reliable model. One such method of training makes use of supervized learning algorithms.

In a small shop two people prepare and sell bread rolls. The space for preparing the bread rolls is small. This space includes a sink and a cash register. Each of the ingredients used to fill the bread rolls is kept in a separate bowl with its own fork or spoon. Sometimes the bowls need to be refilled.

Preparing a bread roll requires the following steps:

• slice the bread roll
• spread its base with some sauce
• add lettuce
• fill with at most two chosen fillings
• cover the roll with the other half
• wrap it in paper.

After that, the roll is ready for payment and collection.

Even when the shop is busy, each person only prepares one bread roll at a time.

A larger restaurant is already using simulation software for the preparation of their dishes.

The restaurant has a menu with eight possible dishes. These dishes may require a number of steps such as cutting, mixing, boiling and grilling to prepare them, as well as different cooking times. The restaurant has a cook and an assistant who prepare one dish at the time.

The rise in global temperatures has led to the melting of the sea ice in polar regions. Scientists have developed a number of computer models that can be used to make predictions about the rate of sea ice melting in these polar regions and its effect on coastal areas globally. In order to create a computer model a number of variables are identified.

A computer model of the effects of the melting of the sea ice in the Arctic Ocean may include the following variables:

• average ocean surface temperature (°C)
• albedo of the ocean (the proportion of light reflected from the ocean surface)
• precipitation (mm)
• salinity of the ocean (grammes of salt in one kilogramme of water)
• area of sea ice (km2).

In this model the following rules have been determined:

• for every 0.01 °C increase in ocean surface temperature, the area of sea ice decreases by 1 %
• for every 1 % decrease in the area of the sea ice, the sea level rises by 20 mm.

The initial values are:

• area of sea ice = 1 000 000 km2
• average surface temperature of the ocean is 0.00 °C.

The scientists observed when running the model numerous times using historical data there were significant differences between observed and expected results.

A second model was developed that included new variables and rules.

The surface of the ocean reflects the heat from the sun. The ratio between the area covered by the sea ice and the area where there is no sea ice (open ocean) affects the value of the average albedo. The lower the albedo, the quicker the sea ice will melt.

The average albedo is calculated using this formula:

Average albedo =

(areaofseaice×albedoofseaice)+(areaofopenocean×albedoofopenocean)(areaofseaice+areaofopenocean)

Note:

• area of sea ice = 1 000 000 km2
• area of open ocean = 1 000 000 km2
• albedo of sea ice = 0.6
• albedo of open ocean = 0.1.

The average albedo will change with every iteration of the model. Each iteration is 2 years after the previous.

The rules for the model are:

• the initial albedo is 0.35
• the rate of decrease in sea ice every 2 years is0.3averagealbedo2
• the starting year is 2019
• the sample rate is every 2 years.

Simulation software can be used to produce a 3D visualization of rising sea levels that change as the user alters the percentage of ice that has melted.

Figure 3: 3D visualization of rising sea levels

[Source: Courtesy NASA/JPL-Caltech.]