{"id":5205,"date":"2024-02-08T12:41:30","date_gmt":"2024-02-08T11:41:30","guid":{"rendered":"https:\/\/ingenius.ecoledesponts.fr\/?p=5205"},"modified":"2024-02-09T17:56:26","modified_gmt":"2024-02-09T16:56:26","slug":"with-artificial-intelligence-operational-research-faces-exciting-technological-industrial-and-environmental-challenges","status":"publish","type":"post","link":"https:\/\/ingenius.ecoledesponts.fr\/en\/articles\/with-artificial-intelligence-operational-research-faces-exciting-technological-industrial-and-environmental-challenges\/","title":{"rendered":"With artificial intelligence, operational research faces \u201cexciting technological, industrial, and environmental challenges.\u201d"},"content":{"rendered":"\n\n\n
\"\"
Ph. Adobe Stock<\/figcaption><\/figure>\n\n\n\n

When you talk about operations research and machine learning, what is the scientific approach you are developing?<\/strong><\/p>\n\n\n\n

Operational research has its roots in military operations and landings. It focuses on the optimal allocation of resources and the optimization of industrial processes. It has three main thrusts. The first very important one, is modeling the industrial process, setting the rules of the game. The second is process industrialization. The greater the volume of operations, the greater the flexibility and the lower the marginal costs per request. The challenge is therefore to write algorithms that can take the entire system and its complexity into account. And finally, the third driver of performance is integrated data processing. One of the emerging approaches is Decision Aware Machine Learning, an area we specialize in at \u00c9cole des Ponts.<\/p>\n\n\n\n

Can you give us an example?<\/strong><\/p>\n\n\n\n

When you order your groceries online, the pickers select your products, put them in bags and then bins. These are then loaded onto a truck that drives around the city. There are two parts to this process: first, the human effort to prepare the required number of bins, and second, an optimization algorithm to manage the delivery rounds. The first part involves predicting a result using statistics and machine learning. The second is a combinatorial optimization problem, because our variables are integers; they are not continuous variables: a whole truck \u2014 not half a truck \u2014 makes the delivery, and it carries whole cartons \u2014 not half cartons.
But the two are linked: to make a profit, the truck has to be optimally loaded. This means higher mileage, but also more deliveries with minimal environmental impact.
Success depends on an accurate estimation. If you overestimate the size of the order, and therefore the number of bins, by even 3%, the truck will only be 95% loaded, which has minimal impact. But if you underestimate it by 3%, the truck will be 101% full, so you will need two trucks.<\/p>\n\n\n\n

In other words, the right model is not simply the one that predicts the right number of bins, but also the one that results in the right number of subsequent deliveries?<\/strong><\/p>\n\n\n\n

Yes, from a mathematical point of view, for maximum performance, you need to take the decisions made during learning into account. In practice, this means treating combinatorial optimization as a layer of a neural network, and merging machine learning and optimization into a single calculation. This enables data-driven programming and decision-making.<\/p>\n\n\n\n

How has your field of research evolved in recent years?<\/strong><\/p>\n\n\n\n

What has changed is the variety of applications we can deploy, because our algorithms are better designed and more refined, and we have massive amounts of qualitative data to draw on. The most obvious advances in our field have been in algorithms. For example, in mixed interger linear programming, if you use the same computer and run the same benchmark on software from 1990 and again on software from 2015, the latter will run 780,000 times faster. And if you factor in the power of computers, the increase in speed is around 450 billion times faster. This means that we have algorithms that scale well, in other words, they can handle a large number of parameters. Today, industry is routinely dealing with integer linear programming problems involving one million variables and one million constraints.<\/p>\n\n\n\n

What are the latest developments?<\/strong><\/p>\n\n\n\n

Data-driven real-time decision-making are something we could not have imagined 10 years ago, but which we have been learning how to do for the last three or four years. The wave of machine learning has made this possible. Deep learning alone is not enough to optimize industrial processes.<\/p>\n\n\n\n

What has changed specifically with machine learning?
Today, we can afford to have tools, purely statistical scientific models that can approximate anything if we calibrate them with the massive amount of data available. And that has an impact on everyone. Before, you needed to know something about statistics; now, you need a background in machine learning. It has become an essential tool, and it answers one of the key questions in science: How do I calibrate my model and reconcile theory with observations?<\/p>\n\n\n\n

What issues do you address with your partners?<\/strong><\/p>\n\n\n\n

The environment, resilience and performance. They are interdependent and need to be addressed simultaneously, as we saw earlier. Imagine that you want to decarbonize your supply chain. You want to move your trucks onto trains? That raises complex optimization issues. What vehicle format should you use? How do you synchronize their arrival with the departure of the trains? And if your fleet is electric, you need to manage not only their range, but also charging requirements and times, while staying on top of price fluctuations. Remember that, when it comes to the environment, there are broadly three levers we can pull. On one side, technological progress, on the other, restrained use, and in the middle, with current technology, rooting out waste: How can I be smarter in my decision-making to avoid wasting human effort, energy, or materials?<\/p>\n\n\n\n

Finally, what does the future hold for your field of research?<\/strong><\/p>\n\n\n\n

In the coming years, the symbiosis of operations research and machine learning, of combinatorial optimization and machine learning will be our guide. In 10 years? We will see. But as I keep telling my students, the technological, industrial, and environmental challenges ahead are fascinating. Right now, one of my PhD students is working on a production startup with an algorithm he built himself. Just imagine that. He is having an impact on Renault\u2019s entire return supply chain \u2013 involving some 1,000 trucks and employees!
Today, the potential for impact and action is tremendous, even within a small team. Algorithms are now everywhere, so research in applied mathematics can have a major industrial impact very quickly. And it can often act as a catalyst for startups. The advantage of doing applied mathematics is that it is the language of science and technology, so you have the privilege of pursuing truly exciting R&D careers.<\/p>\n","protected":false},"excerpt":{"rendered":"

When you talk about operations research and machine learning, what is the scientific approach you are developing? Operational research has […]<\/p>\n","protected":false},"author":6,"featured_media":5132,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_related_content_post":[],"_related_content_subject":[692,690],"_related_content_author":[946],"_related_content_category":[1720],"_related_content_folder":[5213],"_excerpt":"Axel Parmentier has been working at CERMICS (Research Center in Mathematics and Scientific Computing) since 2016. His area of expertise is operations research and machine learning applied to transport, the supply chain, and predictive maintenance. With the growing prevalence of artificial intelligence (AI), the \u00c9cole des Ponts ParisTech expert explains why applied mathematics is poised to play a key role in the future.","_duration":4,"_manual_duration":false,"footnotes":""},"article-types":[27],"class_list":["post-5205","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","article-types-folder"],"has_blocks":true,"block_data":[{"blockName":"enpc\/excerpt","attrs":{"lock":[],"metadata":[],"className":"","style":""},"innerBlocks":[],"innerHTML":"","innerContent":[],"rendered":""},{"blockName":"core\/image","attrs":{"align":"wide","id":5132,"sizeSlug":"large","linkDestination":"none","blob":"","url":"https:\/\/ingenius.ecoledesponts.fr\/wp-content\/uploads\/2024\/01\/AdobeStock_489464055-1024x368.jpeg","alt":"","caption":"Ph. Adobe Stock","lightbox":[],"title":"","href":"","rel":"","linkClass":"","width":"","height":"","aspectRatio":"","scale":"","linkTarget":"","lock":[],"metadata":[],"className":"wp-block-image alignwide size-large","style":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

\"\"
Ph. Adobe Stock<\/figcaption><\/figure>\n","innerContent":["\n
\"\"
Ph. Adobe Stock<\/figcaption><\/figure>\n"],"rendered":"\n
\"\"
Ph. Adobe Stock<\/figcaption><\/figure>\n"},{"blockName":"core\/paragraph","attrs":{"textColor":"red","align":"","content":"When you talk about operations research and machine learning, what is the scientific approach you are developing?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color","style":"","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

When you talk about operations research and machine learning, what is the scientific approach you are developing?<\/strong><\/p>\n","innerContent":["\n

When you talk about operations research and machine learning, what is the scientific approach you are developing?<\/strong><\/p>\n"],"rendered":"\n

When you talk about operations research and machine learning, what is the scientific approach you are developing?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"Operational research has its roots in military operations and landings. It focuses on the optimal allocation of resources and the optimization of industrial processes. It has three main thrusts. The first very important one, is modeling the industrial process, setting the rules of the game. The second is process industrialization. The greater the volume of operations, the greater the flexibility and the lower the marginal costs per request. The challenge is therefore to write algorithms that can take the entire system and its complexity into account. And finally, the third driver of performance is integrated data processing. One of the emerging approaches is Decision Aware Machine Learning, an area we specialize in at \u00c9cole des Ponts.","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

Operational research has its roots in military operations and landings. It focuses on the optimal allocation of resources and the optimization of industrial processes. It has three main thrusts. The first very important one, is modeling the industrial process, setting the rules of the game. The second is process industrialization. The greater the volume of operations, the greater the flexibility and the lower the marginal costs per request. The challenge is therefore to write algorithms that can take the entire system and its complexity into account. And finally, the third driver of performance is integrated data processing. One of the emerging approaches is Decision Aware Machine Learning, an area we specialize in at \u00c9cole des Ponts.<\/p>\n","innerContent":["\n

Operational research has its roots in military operations and landings. It focuses on the optimal allocation of resources and the optimization of industrial processes. It has three main thrusts. The first very important one, is modeling the industrial process, setting the rules of the game. The second is process industrialization. The greater the volume of operations, the greater the flexibility and the lower the marginal costs per request. The challenge is therefore to write algorithms that can take the entire system and its complexity into account. And finally, the third driver of performance is integrated data processing. One of the emerging approaches is Decision Aware Machine Learning, an area we specialize in at \u00c9cole des Ponts.<\/p>\n"],"rendered":"\n

Operational research has its roots in military operations and landings. It focuses on the optimal allocation of resources and the optimization of industrial processes. It has three main thrusts. The first very important one, is modeling the industrial process, setting the rules of the game. The second is process industrialization. The greater the volume of operations, the greater the flexibility and the lower the marginal costs per request. The challenge is therefore to write algorithms that can take the entire system and its complexity into account. And finally, the third driver of performance is integrated data processing. One of the emerging approaches is Decision Aware Machine Learning, an area we specialize in at \u00c9cole des Ponts.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"textColor":"red","align":"","content":"Can you give us an example?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color","style":"","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

Can you give us an example?<\/strong><\/p>\n","innerContent":["\n

Can you give us an example?<\/strong><\/p>\n"],"rendered":"\n

Can you give us an example?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"When you order your groceries online, the pickers select your products, put them in bags and then bins. These are then loaded onto a truck that drives around the city. There are two parts to this process: first, the human effort to prepare the required number of bins, and second, an optimization algorithm to manage the delivery rounds. The first part involves predicting a result using statistics and machine learning. The second is a combinatorial optimization problem, because our variables are integers; they are not continuous variables: a whole truck \u2014 not half a truck \u2014 makes the delivery, and it carries whole cartons \u2014 not half cartons.But the two are linked: to make a profit, the truck has to be optimally loaded. This means higher mileage, but also more deliveries with minimal environmental impact.Success depends on an accurate estimation. If you overestimate the size of the order, and therefore the number of bins, by even 3%, the truck will only be 95% loaded, which has minimal impact. But if you underestimate it by 3%, the truck will be 101% full, so you will need two trucks.","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

When you order your groceries online, the pickers select your products, put them in bags and then bins. These are then loaded onto a truck that drives around the city. There are two parts to this process: first, the human effort to prepare the required number of bins, and second, an optimization algorithm to manage the delivery rounds. The first part involves predicting a result using statistics and machine learning. The second is a combinatorial optimization problem, because our variables are integers; they are not continuous variables: a whole truck \u2014 not half a truck \u2014 makes the delivery, and it carries whole cartons \u2014 not half cartons.
But the two are linked: to make a profit, the truck has to be optimally loaded. This means higher mileage, but also more deliveries with minimal environmental impact.
Success depends on an accurate estimation. If you overestimate the size of the order, and therefore the number of bins, by even 3%, the truck will only be 95% loaded, which has minimal impact. But if you underestimate it by 3%, the truck will be 101% full, so you will need two trucks.<\/p>\n","innerContent":["\n

When you order your groceries online, the pickers select your products, put them in bags and then bins. These are then loaded onto a truck that drives around the city. There are two parts to this process: first, the human effort to prepare the required number of bins, and second, an optimization algorithm to manage the delivery rounds. The first part involves predicting a result using statistics and machine learning. The second is a combinatorial optimization problem, because our variables are integers; they are not continuous variables: a whole truck \u2014 not half a truck \u2014 makes the delivery, and it carries whole cartons \u2014 not half cartons.
But the two are linked: to make a profit, the truck has to be optimally loaded. This means higher mileage, but also more deliveries with minimal environmental impact.
Success depends on an accurate estimation. If you overestimate the size of the order, and therefore the number of bins, by even 3%, the truck will only be 95% loaded, which has minimal impact. But if you underestimate it by 3%, the truck will be 101% full, so you will need two trucks.<\/p>\n"],"rendered":"\n

When you order your groceries online, the pickers select your products, put them in bags and then bins. These are then loaded onto a truck that drives around the city. There are two parts to this process: first, the human effort to prepare the required number of bins, and second, an optimization algorithm to manage the delivery rounds. The first part involves predicting a result using statistics and machine learning. The second is a combinatorial optimization problem, because our variables are integers; they are not continuous variables: a whole truck \u2014 not half a truck \u2014 makes the delivery, and it carries whole cartons \u2014 not half cartons.
But the two are linked: to make a profit, the truck has to be optimally loaded. This means higher mileage, but also more deliveries with minimal environmental impact.
Success depends on an accurate estimation. If you overestimate the size of the order, and therefore the number of bins, by even 3%, the truck will only be 95% loaded, which has minimal impact. But if you underestimate it by 3%, the truck will be 101% full, so you will need two trucks.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"textColor":"red","align":"","content":"In other words, the right model is not simply the one that predicts the right number of bins, but also the one that results in the right number of subsequent deliveries?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color","style":"","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

In other words, the right model is not simply the one that predicts the right number of bins, but also the one that results in the right number of subsequent deliveries?<\/strong><\/p>\n","innerContent":["\n

In other words, the right model is not simply the one that predicts the right number of bins, but also the one that results in the right number of subsequent deliveries?<\/strong><\/p>\n"],"rendered":"\n

In other words, the right model is not simply the one that predicts the right number of bins, but also the one that results in the right number of subsequent deliveries?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"Yes, from a mathematical point of view, for maximum performance, you need to take the decisions made during learning into account. In practice, this means treating combinatorial optimization as a layer of a neural network, and merging machine learning and optimization into a single calculation. This enables data-driven programming and decision-making.","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

Yes, from a mathematical point of view, for maximum performance, you need to take the decisions made during learning into account. In practice, this means treating combinatorial optimization as a layer of a neural network, and merging machine learning and optimization into a single calculation. This enables data-driven programming and decision-making.<\/p>\n","innerContent":["\n

Yes, from a mathematical point of view, for maximum performance, you need to take the decisions made during learning into account. In practice, this means treating combinatorial optimization as a layer of a neural network, and merging machine learning and optimization into a single calculation. This enables data-driven programming and decision-making.<\/p>\n"],"rendered":"\n

Yes, from a mathematical point of view, for maximum performance, you need to take the decisions made during learning into account. In practice, this means treating combinatorial optimization as a layer of a neural network, and merging machine learning and optimization into a single calculation. This enables data-driven programming and decision-making.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"textColor":"red","align":"","content":"How has your field of research evolved in recent years?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color","style":"","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

How has your field of research evolved in recent years?<\/strong><\/p>\n","innerContent":["\n

How has your field of research evolved in recent years?<\/strong><\/p>\n"],"rendered":"\n

How has your field of research evolved in recent years?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"What has changed is the variety of applications we can deploy, because our algorithms are better designed and more refined, and we have massive amounts of qualitative data to draw on. The most obvious advances in our field have been in algorithms. For example, in mixed interger linear programming, if you use the same computer and run the same benchmark on software from 1990 and again on software from 2015, the latter will run 780,000 times faster. And if you factor in the power of computers, the increase in speed is around 450 billion times faster. This means that we have algorithms that scale well, in other words, they can handle a large number of parameters. Today, industry is routinely dealing with integer linear programming problems involving one million variables and one million constraints.","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

What has changed is the variety of applications we can deploy, because our algorithms are better designed and more refined, and we have massive amounts of qualitative data to draw on. The most obvious advances in our field have been in algorithms. For example, in mixed interger linear programming, if you use the same computer and run the same benchmark on software from 1990 and again on software from 2015, the latter will run 780,000 times faster. And if you factor in the power of computers, the increase in speed is around 450 billion times faster. This means that we have algorithms that scale well, in other words, they can handle a large number of parameters. Today, industry is routinely dealing with integer linear programming problems involving one million variables and one million constraints.<\/p>\n","innerContent":["\n

What has changed is the variety of applications we can deploy, because our algorithms are better designed and more refined, and we have massive amounts of qualitative data to draw on. The most obvious advances in our field have been in algorithms. For example, in mixed interger linear programming, if you use the same computer and run the same benchmark on software from 1990 and again on software from 2015, the latter will run 780,000 times faster. And if you factor in the power of computers, the increase in speed is around 450 billion times faster. This means that we have algorithms that scale well, in other words, they can handle a large number of parameters. Today, industry is routinely dealing with integer linear programming problems involving one million variables and one million constraints.<\/p>\n"],"rendered":"\n

What has changed is the variety of applications we can deploy, because our algorithms are better designed and more refined, and we have massive amounts of qualitative data to draw on. The most obvious advances in our field have been in algorithms. For example, in mixed interger linear programming, if you use the same computer and run the same benchmark on software from 1990 and again on software from 2015, the latter will run 780,000 times faster. And if you factor in the power of computers, the increase in speed is around 450 billion times faster. This means that we have algorithms that scale well, in other words, they can handle a large number of parameters. Today, industry is routinely dealing with integer linear programming problems involving one million variables and one million constraints.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"textColor":"red","align":"","content":"What are the latest developments?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color","style":"","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

What are the latest developments?<\/strong><\/p>\n","innerContent":["\n

What are the latest developments?<\/strong><\/p>\n"],"rendered":"\n

What are the latest developments?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"Data-driven real-time decision-making are something we could not have imagined 10 years ago, but which we have been learning how to do for the last three or four years. The wave of machine learning has made this possible. Deep learning alone is not enough to optimize industrial processes.","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

Data-driven real-time decision-making are something we could not have imagined 10 years ago, but which we have been learning how to do for the last three or four years. The wave of machine learning has made this possible. Deep learning alone is not enough to optimize industrial processes.<\/p>\n","innerContent":["\n

Data-driven real-time decision-making are something we could not have imagined 10 years ago, but which we have been learning how to do for the last three or four years. The wave of machine learning has made this possible. Deep learning alone is not enough to optimize industrial processes.<\/p>\n"],"rendered":"\n

Data-driven real-time decision-making are something we could not have imagined 10 years ago, but which we have been learning how to do for the last three or four years. The wave of machine learning has made this possible. Deep learning alone is not enough to optimize industrial processes.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"What has changed specifically with machine learning?Today, we can afford to have tools, purely statistical scientific models that can approximate anything if we calibrate them with the massive amount of data available. And that has an impact on everyone. Before, you needed to know something about statistics; now, you need a background in machine learning. It has become an essential tool, and it answers one of the key questions in science: How do I calibrate my model and reconcile theory with observations?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

What has changed specifically with machine learning?
Today, we can afford to have tools, purely statistical scientific models that can approximate anything if we calibrate them with the massive amount of data available. And that has an impact on everyone. Before, you needed to know something about statistics; now, you need a background in machine learning. It has become an essential tool, and it answers one of the key questions in science: How do I calibrate my model and reconcile theory with observations?<\/p>\n","innerContent":["\n

What has changed specifically with machine learning?
Today, we can afford to have tools, purely statistical scientific models that can approximate anything if we calibrate them with the massive amount of data available. And that has an impact on everyone. Before, you needed to know something about statistics; now, you need a background in machine learning. It has become an essential tool, and it answers one of the key questions in science: How do I calibrate my model and reconcile theory with observations?<\/p>\n"],"rendered":"\n

What has changed specifically with machine learning?
Today, we can afford to have tools, purely statistical scientific models that can approximate anything if we calibrate them with the massive amount of data available. And that has an impact on everyone. Before, you needed to know something about statistics; now, you need a background in machine learning. It has become an essential tool, and it answers one of the key questions in science: How do I calibrate my model and reconcile theory with observations?<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"textColor":"red","align":"","content":"What issues do you address with your partners?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color","style":"","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

What issues do you address with your partners?<\/strong><\/p>\n","innerContent":["\n

What issues do you address with your partners?<\/strong><\/p>\n"],"rendered":"\n

What issues do you address with your partners?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"The environment, resilience and performance. They are interdependent and need to be addressed simultaneously, as we saw earlier. Imagine that you want to decarbonize your supply chain. You want to move your trucks onto trains? That raises complex optimization issues. What vehicle format should you use? How do you synchronize their arrival with the departure of the trains? And if your fleet is electric, you need to manage not only their range, but also charging requirements and times, while staying on top of price fluctuations. Remember that, when it comes to the environment, there are broadly three levers we can pull. On one side, technological progress, on the other, restrained use, and in the middle, with current technology, rooting out waste: How can I be smarter in my decision-making to avoid wasting human effort, energy, or materials?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

The environment, resilience and performance. They are interdependent and need to be addressed simultaneously, as we saw earlier. Imagine that you want to decarbonize your supply chain. You want to move your trucks onto trains? That raises complex optimization issues. What vehicle format should you use? How do you synchronize their arrival with the departure of the trains? And if your fleet is electric, you need to manage not only their range, but also charging requirements and times, while staying on top of price fluctuations. Remember that, when it comes to the environment, there are broadly three levers we can pull. On one side, technological progress, on the other, restrained use, and in the middle, with current technology, rooting out waste: How can I be smarter in my decision-making to avoid wasting human effort, energy, or materials?<\/p>\n","innerContent":["\n

The environment, resilience and performance. They are interdependent and need to be addressed simultaneously, as we saw earlier. Imagine that you want to decarbonize your supply chain. You want to move your trucks onto trains? That raises complex optimization issues. What vehicle format should you use? How do you synchronize their arrival with the departure of the trains? And if your fleet is electric, you need to manage not only their range, but also charging requirements and times, while staying on top of price fluctuations. Remember that, when it comes to the environment, there are broadly three levers we can pull. On one side, technological progress, on the other, restrained use, and in the middle, with current technology, rooting out waste: How can I be smarter in my decision-making to avoid wasting human effort, energy, or materials?<\/p>\n"],"rendered":"\n

The environment, resilience and performance. They are interdependent and need to be addressed simultaneously, as we saw earlier. Imagine that you want to decarbonize your supply chain. You want to move your trucks onto trains? That raises complex optimization issues. What vehicle format should you use? How do you synchronize their arrival with the departure of the trains? And if your fleet is electric, you need to manage not only their range, but also charging requirements and times, while staying on top of price fluctuations. Remember that, when it comes to the environment, there are broadly three levers we can pull. On one side, technological progress, on the other, restrained use, and in the middle, with current technology, rooting out waste: How can I be smarter in my decision-making to avoid wasting human effort, energy, or materials?<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"textColor":"red","align":"","content":"Finally, what does the future hold for your field of research?","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color","style":"","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

Finally, what does the future hold for your field of research?<\/strong><\/p>\n","innerContent":["\n

Finally, what does the future hold for your field of research?<\/strong><\/p>\n"],"rendered":"\n

Finally, what does the future hold for your field of research?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":"In the coming years, the symbiosis of operations research and machine learning, of combinatorial optimization and machine learning will be our guide. In 10 years? We will see. But as I keep telling my students, the technological, industrial, and environmental challenges ahead are fascinating. Right now, one of my PhD students is working on a production startup with an algorithm he built himself. Just imagine that. He is having an impact on Renault\u2019s entire return supply chain \u2013 involving some 1,000 trucks and employees!Today, the potential for impact and action is tremendous, even within a small team. Algorithms are now everywhere, so research in applied mathematics can have a major industrial impact very quickly. And it can often act as a catalyst for startups. The advantage of doing applied mathematics is that it is the language of science and technology, so you have the privilege of pursuing truly exciting R&D careers.","dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

In the coming years, the symbiosis of operations research and machine learning, of combinatorial optimization and machine learning will be our guide. In 10 years? We will see. But as I keep telling my students, the technological, industrial, and environmental challenges ahead are fascinating. Right now, one of my PhD students is working on a production startup with an algorithm he built himself. Just imagine that. He is having an impact on Renault\u2019s entire return supply chain \u2013 involving some 1,000 trucks and employees!
Today, the potential for impact and action is tremendous, even within a small team. Algorithms are now everywhere, so research in applied mathematics can have a major industrial impact very quickly. And it can often act as a catalyst for startups. The advantage of doing applied mathematics is that it is the language of science and technology, so you have the privilege of pursuing truly exciting R&D careers.<\/p>\n","innerContent":["\n

In the coming years, the symbiosis of operations research and machine learning, of combinatorial optimization and machine learning will be our guide. In 10 years? We will see. But as I keep telling my students, the technological, industrial, and environmental challenges ahead are fascinating. Right now, one of my PhD students is working on a production startup with an algorithm he built himself. Just imagine that. He is having an impact on Renault\u2019s entire return supply chain \u2013 involving some 1,000 trucks and employees!
Today, the potential for impact and action is tremendous, even within a small team. Algorithms are now everywhere, so research in applied mathematics can have a major industrial impact very quickly. And it can often act as a catalyst for startups. The advantage of doing applied mathematics is that it is the language of science and technology, so you have the privilege of pursuing truly exciting R&D careers.<\/p>\n"],"rendered":"\n

In the coming years, the symbiosis of operations research and machine learning, of combinatorial optimization and machine learning will be our guide. In 10 years? We will see. But as I keep telling my students, the technological, industrial, and environmental challenges ahead are fascinating. Right now, one of my PhD students is working on a production startup with an algorithm he built himself. Just imagine that. He is having an impact on Renault\u2019s entire return supply chain \u2013 involving some 1,000 trucks and employees!
Today, the potential for impact and action is tremendous, even within a small team. Algorithms are now everywhere, so research in applied mathematics can have a major industrial impact very quickly. And it can often act as a catalyst for startups. The advantage of doing applied mathematics is that it is the language of science and technology, so you have the privilege of pursuing truly exciting R&D careers.<\/p>\n"}],"seo":{"title":"With artificial intelligence, operational research faces \u201cexciting technological, industrial, and environmental challenges.\u201d"},"media":{"img":"\"\"","src":"https:\/\/ingenius.ecoledesponts.fr\/wp-content\/uploads\/2024\/01\/AdobeStock_489464055-scaled.jpeg"},"url":"\/en\/articles\/with-artificial-intelligence-operational-research-faces-exciting-technological-industrial-and-environmental-challenges\/","related":{"post":[],"author":[{"title":"Axel Parmentier","url":"\/en\/authors\/axel-parmentier\/","id":"946","media":"\"\"","slug":"axel-parmentier"}],"subject":[{"title":"Mobility, Transport & Infrastructure","url":"\/en\/subjects\/mobility-transport-infrastructure\/","id":"692","media":"\"\"","slug":"mobility-transport-infrastructure"},{"title":"Digital Technology, Modeling & Artificial Intelligence","url":"\/en\/subjects\/digital-technology-modeling-artificial-intelligence\/","id":"690","media":"\"\"","slug":"digital-technology-modeling-artificial-intelligence"}],"category":[{"title":"Article collection","url":"\/en\/articles\/category\/dossier\/","id":"1720","media":"","slug":"dossier","_related_post_type":"folder"}],"folder":[{"title":"Artificial intelligence for engineering","url":"\/en\/folders\/artificial-intelligence-for-engineering\/","id":"5213","media":"\"\"","slug":"artificial-intelligence-for-engineering"}]},"translated":"https:\/\/ingenius.ecoledesponts.fr\/articles\/la-recherche-operationnelle\/","icon":"icon-folder","duration":"4","custom_excerpt":"Axel Parmentier has been working at CERMICS (Research Center in Mathematics and Scientific Computing) since 2016. His area of expertise is operations research and machine learning applied to transport, the supply chain, and predictive maintenance. With the growing prevalence of artificial intelligence (AI), the \u00c9cole des Ponts ParisTech expert explains why applied mathematics is poised to play a key role in the future.","duration_type":"","_links":{"self":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts\/5205","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/users\/6"}],"replies":[{"embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/comments?post=5205"}],"version-history":[{"count":2,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts\/5205\/revisions"}],"predecessor-version":[{"id":5208,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts\/5205\/revisions\/5208"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/media\/5132"}],"wp:attachment":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/media?parent=5205"}],"wp:term":[{"taxonomy":"article-types","embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/article-types?post=5205"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}