In this work we introduce a differentiable version of the Compositional Pattern Producing Network... more In this work we introduce a differentiable version of the Compositional Pattern Producing Network, called the DPPN. Unlike a standard CPPN, the topology of a DPPN is evolved but the weights are learned. A Lamarckian algorithm, that combines evolution and learning, produces DPPNs to reconstruct an image. Our main result is that DPPNs can be evolved/trained to compress the weights of a denoising autoencoder from 157684 to roughly 200 parameters, while achieving a reconstruction accuracy comparable to a fully connected network with more than two orders of magnitude more parameters. The regularization ability of the DPPN allows it to rediscover (approximate) convolutional network architectures embedded within a fully connected architecture. Such convolutional architectures are the current state of the art for many computer vision applications, so it is satisfying that DPPNs are capable of discovering this structure rather than having to build it in by design. DPPNs exhibit better genera...
Whilst there are perhaps only a few scientific methods, there seem to be almost as many artistic ... more Whilst there are perhaps only a few scientific methods, there seem to be almost as many artistic methods as there are artists. Artistic processes appear to inhabit the highest order of open-endedness. To begin to understand some of the processes of art making it is helpful to try to automate them even partially. In this paper, a novel algorithm for producing generative art is described which allows a user to input a text string, and which in a creative response to this string, outputs an image which interprets that string. It does so by evolving images using a hierarchical neural Lindenmeyer system, and evaluating these images along the way using an image text dual encoder trained on billions of images and their associated text from the internet. In doing so we have access to and control over an instance of an artistic process, allowing analysis of which aspects of the artistic process become the task of the algorithm, and which elements remain the responsibility of the artist.
Brad Alexander Juan Herrero Justyna Petke Shaukat Ali Malcolm Heywood Stjepan Picek Andrea Arcuri... more Brad Alexander Juan Herrero Justyna Petke Shaukat Ali Malcolm Heywood Stjepan Picek Andrea Arcuri Cezary Z. Janikow Nelishia Pillay Ignacio Arnaldo Colin Johnson John Robinson R. Muhammad Atif Azad Anna Jordanous Patricia Ryser-Welch Dylan Banarse Paul Kaufmann Yago Saez Amit Benbassat Edward Keedwell Ivan Sekaj Peter Bentley Krzysztof Krawiec Kisung Seo Michal Bidlo Pavel Kromer Sara Silva Stefano Cagnoni Stuart Lacy Kevin Sim Jeffrey Chan William Langdon James Smith Francis Chicano Xianneng Li Andy Song Vic Ciesielski Luca Manzoni Giovanni Squillero German Creamer James McDermott Kenneth Stanley Márjory Da Costa-Abreu Yi Mei Thomas Stutzle Roland Dobai Maizura Mokhtar Petr Svenda Alan Dorin Kourosh Neshatian Jerry Swan Liang Gao Su Nguyen Marcin Szubert Oscar Garnica Trung Thanh Nguyen Gianluca Tempesti Mario Giacobini Randy Olson Hsing-Chih Tsai
Reinforcement learning (RL) has proven to be a powerful paradigm for deriving complex behaviors f... more Reinforcement learning (RL) has proven to be a powerful paradigm for deriving complex behaviors from simple reward signals in a wide range of environments. When applying RL to continuous control agents in simulated physics environments, the body is usually considered to be part of the environment. However, during evolution the physical body of biological organisms and their controlling brains are co-evolved, thus exploring a much larger space of actuator/controller configurations. Put differently, the intelligence does not reside only in the agent's mind, but also in the design of their body. We propose a method for uncovering strong agents, consisting of a good combination of a body and policy, based on combining RL with an evolutionary procedure. Given the resulting agent, we also propose an approach for identifying the body changes that contributed the most to the agent performance. We use the Shapley value from cooperative game theory to find the fair contribution of individ...
For artificial general intelligence (AGI) it would be efficient if multiple users trained the sam... more For artificial general intelligence (AGI) it would be efficient if multiple users trained the same giant neural network, permitting parameter reuse, without catastrophic forgetting. PathNet is a first step in this direction. It is a neural network algorithm that uses agents embedded in the neural network whose task is to discover which parts of the network to re-use for new tasks. Agents are pathways (views) through the network which determine the subset of parameters that are used and updated by the forwards and backwards passes of the backpropogation algorithm. During learning, a tournament selection genetic algorithm is used to select pathways through the neural network for replication and mutation. Pathway fitness is the performance of that pathway measured according to a cost function. We demonstrate successful transfer learning; fixing the parameters along a path learned on task A and re-evolving a new population of paths for task B, allows task B to be learned faster than it ...
This paper describes a neural network architecture that has been developed to perform deformation... more This paper describes a neural network architecture that has been developed to perform deformation tolerant object recognition from grey-scale images. It uses a form of deformable template matching, generating new templates in a self-organising manner. The results demonstrate the network's ability to build classes when no suitable classes are available. The amount of deformation allowed within a class can be controlled to allow the network to be applied to a wide range of applications. Results are presented for a set of generated images which allow the effects of the selection of the major network parameters to be shown.
... the network is forced to create a new class. With the use of hypothesis testing the weight,s ... more ... the network is forced to create a new class. With the use of hypothesis testing the weight,s of the PDMs are cornpared tro t,lie features extracted from the presented image. The amount, of mismatch dlowed is controlled by tlie ...
In this work we introduce a differentiable version of the Compositional Pattern Producing Network... more In this work we introduce a differentiable version of the Compositional Pattern Producing Network, called the DPPN. Unlike a standard CPPN, the topology of a DPPN is evolved but the weights are learned. A Lamarckian algorithm, that combines evolution and learning, produces DPPNs to reconstruct an image. Our main result is that DPPNs can be evolved/trained to compress the weights of a denoising autoencoder from 157684 to roughly 200 parameters, while achieving a reconstruction accuracy comparable to a fully connected network with more than two orders of magnitude more parameters. The regularization ability of the DPPN allows it to rediscover (approximate) convolutional network architectures embedded within a fully connected architecture. Such convolutional architectures are the current state of the art for many computer vision applications, so it is satisfying that DPPNs are capable of discovering this structure rather than having to build it in by design. DPPNs exhibit better genera...
Whilst there are perhaps only a few scientific methods, there seem to be almost as many artistic ... more Whilst there are perhaps only a few scientific methods, there seem to be almost as many artistic methods as there are artists. Artistic processes appear to inhabit the highest order of open-endedness. To begin to understand some of the processes of art making it is helpful to try to automate them even partially. In this paper, a novel algorithm for producing generative art is described which allows a user to input a text string, and which in a creative response to this string, outputs an image which interprets that string. It does so by evolving images using a hierarchical neural Lindenmeyer system, and evaluating these images along the way using an image text dual encoder trained on billions of images and their associated text from the internet. In doing so we have access to and control over an instance of an artistic process, allowing analysis of which aspects of the artistic process become the task of the algorithm, and which elements remain the responsibility of the artist.
Brad Alexander Juan Herrero Justyna Petke Shaukat Ali Malcolm Heywood Stjepan Picek Andrea Arcuri... more Brad Alexander Juan Herrero Justyna Petke Shaukat Ali Malcolm Heywood Stjepan Picek Andrea Arcuri Cezary Z. Janikow Nelishia Pillay Ignacio Arnaldo Colin Johnson John Robinson R. Muhammad Atif Azad Anna Jordanous Patricia Ryser-Welch Dylan Banarse Paul Kaufmann Yago Saez Amit Benbassat Edward Keedwell Ivan Sekaj Peter Bentley Krzysztof Krawiec Kisung Seo Michal Bidlo Pavel Kromer Sara Silva Stefano Cagnoni Stuart Lacy Kevin Sim Jeffrey Chan William Langdon James Smith Francis Chicano Xianneng Li Andy Song Vic Ciesielski Luca Manzoni Giovanni Squillero German Creamer James McDermott Kenneth Stanley Márjory Da Costa-Abreu Yi Mei Thomas Stutzle Roland Dobai Maizura Mokhtar Petr Svenda Alan Dorin Kourosh Neshatian Jerry Swan Liang Gao Su Nguyen Marcin Szubert Oscar Garnica Trung Thanh Nguyen Gianluca Tempesti Mario Giacobini Randy Olson Hsing-Chih Tsai
Reinforcement learning (RL) has proven to be a powerful paradigm for deriving complex behaviors f... more Reinforcement learning (RL) has proven to be a powerful paradigm for deriving complex behaviors from simple reward signals in a wide range of environments. When applying RL to continuous control agents in simulated physics environments, the body is usually considered to be part of the environment. However, during evolution the physical body of biological organisms and their controlling brains are co-evolved, thus exploring a much larger space of actuator/controller configurations. Put differently, the intelligence does not reside only in the agent's mind, but also in the design of their body. We propose a method for uncovering strong agents, consisting of a good combination of a body and policy, based on combining RL with an evolutionary procedure. Given the resulting agent, we also propose an approach for identifying the body changes that contributed the most to the agent performance. We use the Shapley value from cooperative game theory to find the fair contribution of individ...
For artificial general intelligence (AGI) it would be efficient if multiple users trained the sam... more For artificial general intelligence (AGI) it would be efficient if multiple users trained the same giant neural network, permitting parameter reuse, without catastrophic forgetting. PathNet is a first step in this direction. It is a neural network algorithm that uses agents embedded in the neural network whose task is to discover which parts of the network to re-use for new tasks. Agents are pathways (views) through the network which determine the subset of parameters that are used and updated by the forwards and backwards passes of the backpropogation algorithm. During learning, a tournament selection genetic algorithm is used to select pathways through the neural network for replication and mutation. Pathway fitness is the performance of that pathway measured according to a cost function. We demonstrate successful transfer learning; fixing the parameters along a path learned on task A and re-evolving a new population of paths for task B, allows task B to be learned faster than it ...
This paper describes a neural network architecture that has been developed to perform deformation... more This paper describes a neural network architecture that has been developed to perform deformation tolerant object recognition from grey-scale images. It uses a form of deformable template matching, generating new templates in a self-organising manner. The results demonstrate the network's ability to build classes when no suitable classes are available. The amount of deformation allowed within a class can be controlled to allow the network to be applied to a wide range of applications. Results are presented for a set of generated images which allow the effects of the selection of the major network parameters to be shown.
... the network is forced to create a new class. With the use of hypothesis testing the weight,s ... more ... the network is forced to create a new class. With the use of hypothesis testing the weight,s of the PDMs are cornpared tro t,lie features extracted from the presented image. The amount, of mismatch dlowed is controlled by tlie ...
Uploads
Papers by Dylan Banarse