Generative Tensorial Reinforcement Learning (GENTRL) - Revision history

BPeat at 01:12, 10 April 2024

2024-04-10T01:12:48Z

BPeat: Text replacement - "* Conversational AI ... ChatGPT | OpenAI ... Bing | Microsoft ... Bard | Google ... Claude | Anthropic ... Perplexity ... You ... Ernie | Baidu" to "* Conversational AI ... [[C...

2024-03-16T15:48:36Z

BPeat at 18:51, 16 September 2023

2023-09-16T18:51:48Z

BPeat at 18:51, 16 September 2023

2023-09-16T18:51:04Z

BPeat at 18:50, 16 September 2023

2023-09-16T18:50:14Z

BPeat at 00:42, 3 September 2023

2023-09-03T00:42:01Z

BPeat at 00:18, 13 July 2023

2023-07-13T00:18:41Z

BPeat at 17:15, 15 April 2023

2023-04-15T17:15:01Z

BPeat: Text replacement - "BingAI" to "Bing"

2023-03-31T01:39:54Z

Text replacement - "BingAI" to "Bing"

BPeat: Text replacement - "http:" to "https:"

2023-03-28T19:51:26Z

Text replacement - "http:" to "https:"

@@ Line 17: / Line 17: @@
 * [[Pharmaceuticals]]
 * [[What is Artificial Intelligence (AI)? | Artificial Intelligence (AI)]] ... [[Generative AI]] ... [[Machine Learning (ML)]] ... [[Deep Learning]] ... [[Neural Network]] ... [[Reinforcement Learning (RL)|Reinforcement]] ... [[Learning Techniques]]
-* [[Conversational AI]] ... [[ChatGPT]] | [[OpenAI]] ... [[Bing/Copilot]] | [[Microsoft]] ... [[Gemini]] | [[Google]] ... [[Claude]] | [[Anthropic]] ... [[Perplexity]] ... [[You]] ... [[phind]] ... [[Ernie]] | [[Baidu]]
+* [[Conversational AI]] ... [[ChatGPT]] | [[OpenAI]] ... [[Bing/Copilot]] | [[Microsoft]] ... [[Gemini]] | [[Google]] ... [[Claude]] | [[Anthropic]] ... [[Perplexity]] ... [[You]] ... [[phind]] ... [[Grok]] | [https://x.ai/ xAI] ... [[Groq]] ... [[Ernie]] | [[Baidu]]

@@ Line 17: / Line 17: @@
 * [[Pharmaceuticals]]
 * [[What is Artificial Intelligence (AI)? | Artificial Intelligence (AI)]] ... [[Generative AI]] ... [[Machine Learning (ML)]] ... [[Deep Learning]] ... [[Neural Network]] ... [[Reinforcement Learning (RL)|Reinforcement]] ... [[Learning Techniques]]
-* [[Conversational AI]] ... [[ChatGPT]] | [[OpenAI]] ... [[Bing]] | [[Microsoft]] ... [[Bard]] | [[Google]] ... [[Claude]] | [[Anthropic]] ... [[Perplexity]] ... [[You]] ... [[Ernie]] | [[Baidu]]
+* [[Conversational AI]] ... [[ChatGPT]] | [[OpenAI]] ... [[Bing/Copilot]] | [[Microsoft]] ... [[Gemini]] | [[Google]] ... [[Claude]] | [[Anthropic]] ... [[Perplexity]] ... [[You]] ... [[phind]] ... [[Ernie]] | [[Baidu]]

← Older revision		Revision as of 18:51, 16 September 2023
Line 20:		Line 20:


−	A deep [[Generative AI\|generative]] model, [[Generative AI\|generative]] tensorial [[Reinforcement Learning (RL)\|reinforcement learning]] (GENTRL). The GENTRL model is a [[Variational Autoencoder (VAE)]] with a rich prior distribution of the [[latent]] space. We used tensor decompositions to encode the relations between molecular structures and their properties and to learn on data with missing values. We train the model in two steps. First, we learn a mapping of a chemical space on the [[latent]] manifold by maximizing the evidence lower bound. We then freeze all the parameters except for the learnable prior and explore the chemical space to find molecules with a high reward. ~~Generative Tensorial Reinforcement Learning (~~GENTRL) is a deep generative model that can be used for de novo drug discovery and other tasks that require the generation of new molecules with desired properties. GENTRL is a variational autoencoder (VAE) with a rich prior distribution of the [[latent]] space. This allows GENTRL to generate a diverse range of molecules with varying properties.	+	A deep [[Generative AI\|generative]] model, [[Generative AI\|generative]] tensorial [[Reinforcement Learning (RL)\|reinforcement learning]] (GENTRL). The GENTRL model is a [[Variational Autoencoder (VAE)]] with a rich prior distribution of the [[latent]] space. We used tensor decompositions to encode the relations between molecular structures and their properties and to learn on data with missing values. We train the model in two steps. First, we learn a mapping of a chemical space on the [[latent]] manifold by maximizing the evidence lower bound. We then freeze all the parameters except for the learnable prior and explore the chemical space to find molecules with a high reward. GENTRL is a deep generative model that can be used for de novo drug discovery and other tasks that require the generation of new molecules with desired properties. GENTRL is a variational autoencoder (VAE) with a rich prior distribution of the [[latent]] space. This allows GENTRL to generate a diverse range of molecules with varying properties.

	GENTRL is trained to optimize three objectives: synthetic feasibility, novelty, and biological activity. Synthetic feasibility is the likelihood that a molecule can be synthesized in a laboratory. Novelty is the likelihood that a molecule is new and has not been previously synthesized. Biological activity is the likelihood that a molecule has the desired biological effect.		GENTRL is trained to optimize three objectives: synthetic feasibility, novelty, and biological activity. Synthetic feasibility is the likelihood that a molecule can be synthesized in a laboratory. Novelty is the likelihood that a molecule is new and has not been previously synthesized. Biological activity is the likelihood that a molecule has the desired biological effect.

← Older revision		Revision as of 18:51, 16 September 2023
Line 20:		Line 20:


−	A deep [[Generative AI\|generative]] model, [[Generative AI\|generative]] tensorial [[Reinforcement Learning (RL)\|reinforcement learning]] (GENTRL). The GENTRL model is a [[Variational Autoencoder (VAE)]] with a rich prior distribution of the latent space. We used tensor decompositions to encode the relations between molecular structures and their properties and to learn on data with missing values. We train the model in two steps. First, we learn a mapping of a chemical space on the latent manifold by maximizing the evidence lower bound. We then freeze all the parameters except for the learnable prior and explore the chemical space to find molecules with a high reward. Generative Tensorial Reinforcement Learning (GENTRL) is a deep generative model that can be used for de novo drug discovery and other tasks that require the generation of new molecules with desired properties. GENTRL is a variational autoencoder (VAE) with a rich prior distribution of the latent space. This allows GENTRL to generate a diverse range of molecules with varying properties.	+	A deep [[Generative AI\|generative]] model, [[Generative AI\|generative]] tensorial [[Reinforcement Learning (RL)\|reinforcement learning]] (GENTRL). The GENTRL model is a [[Variational Autoencoder (VAE)]] with a rich prior distribution of the [[latent]] space. We used tensor decompositions to encode the relations between molecular structures and their properties and to learn on data with missing values. We train the model in two steps. First, we learn a mapping of a chemical space on the [[latent]] manifold by maximizing the evidence lower bound. We then freeze all the parameters except for the learnable prior and explore the chemical space to find molecules with a high reward. Generative Tensorial Reinforcement Learning (GENTRL) is a deep generative model that can be used for de novo drug discovery and other tasks that require the generation of new molecules with desired properties. GENTRL is a variational autoencoder (VAE) with a rich prior distribution of the [[latent]] space. This allows GENTRL to generate a diverse range of molecules with varying properties.

	GENTRL is trained to optimize three objectives: synthetic feasibility, novelty, and biological activity. Synthetic feasibility is the likelihood that a molecule can be synthesized in a laboratory. Novelty is the likelihood that a molecule is new and has not been previously synthesized. Biological activity is the likelihood that a molecule has the desired biological effect.		GENTRL is trained to optimize three objectives: synthetic feasibility, novelty, and biological activity. Synthetic feasibility is the likelihood that a molecule can be synthesized in a laboratory. Novelty is the likelihood that a molecule is new and has not been previously synthesized. Biological activity is the likelihood that a molecule has the desired biological effect.

← Older revision		Revision as of 18:50, 16 September 2023
Line 20:		Line 20:


−	A deep [[Generative AI\|generative]] model, [[Generative AI\|generative]] tensorial [[Reinforcement Learning (RL)\|reinforcement learning]] (GENTRL). The GENTRL model is a [[Variational Autoencoder (VAE)]] with a rich prior distribution of the latent space. We used tensor decompositions to encode the relations between molecular structures and their properties and to learn on data with missing values. We train the model in two steps. First, we learn a mapping of a chemical space on the latent manifold by maximizing the evidence lower bound. We then freeze all the parameters except for the learnable prior and explore the chemical space to find molecules with a high reward.	+	A deep [[Generative AI\|generative]] model, [[Generative AI\|generative]] tensorial [[Reinforcement Learning (RL)\|reinforcement learning]] (GENTRL). The GENTRL model is a [[Variational Autoencoder (VAE)]] with a rich prior distribution of the latent space. We used tensor decompositions to encode the relations between molecular structures and their properties and to learn on data with missing values. We train the model in two steps. First, we learn a mapping of a chemical space on the latent manifold by maximizing the evidence lower bound. We then freeze all the parameters except for the learnable prior and explore the chemical space to find molecules with a high reward. Generative Tensorial Reinforcement Learning (GENTRL) is a deep generative model that can be used for de novo drug discovery and other tasks that require the generation of new molecules with desired properties. GENTRL is a variational autoencoder (VAE) with a rich prior distribution of the latent space. This allows GENTRL to generate a diverse range of molecules with varying properties.
		+
		+	GENTRL is trained to optimize three objectives: synthetic feasibility, novelty, and biological activity. Synthetic feasibility is the likelihood that a molecule can be synthesized in a laboratory. Novelty is the likelihood that a molecule is new and has not been previously synthesized. Biological activity is the likelihood that a molecule has the desired biological effect.
		+
		+	GENTRL is trained in two steps. First, a VAE is trained to learn the distribution of known molecules. This is done by maximizing the evidence lower bound (ELBO). Second, a reinforcement learning (RL) agent is trained to generate molecules that optimize the three objectives. The RL agent is rewarded for generating molecules that are synthetically feasible, novel, and have the desired biological activity.
		+
		+	GENTRL has been shown to be effective in discovering new drug candidates. For example, GENTRL was used to discover potent inhibitors of discoidin domain receptor 1 (DDR1), a kinase target implicated in fibrosis and other diseases. In 21 days, GENTRL was able to discover four compounds that were active in biochemical assays and two compounds that were validated in cell-based assays. One lead candidate was tested and demonstrated favorable pharmacokinetics in mice.
		+
		+	GENTRL is a promising new approach to de novo drug discovery and other tasks that require the generation of new molecules with desired properties. GENTRL is able to generate a diverse range of molecules with high synthetic feasibility, novelty, and biological activity.
		+
		+	Here is an example of how GENTRL can be used to discover new drug candidates:
		+
		+	# A GENTRL model is trained on a dataset of known molecules, including their biological activity.
		+	# The GENTRL model is used to generate a new molecule with the desired biological activity.
		+	# The new molecule is synthesized in a laboratory and tested for its biological activity.
		+	# If the new molecule has the desired biological activity, it is further tested and evaluated for its safety and efficacy.
		+
		+	This process can be repeated to generate and test new drug candidates until a promising lead compound is identified.
		+
		+	GENTRL is a powerful tool for drug discovery and other tasks that require the generation of new molecules with desired properties. GENTRL is still under development, but it has the potential to revolutionize the way that new drugs are discovered.

@@ Line 16: / Line 16: @@
 * [[Generative Adversarial Network (GAN)]]
 * [[Pharmaceuticals]]
-* [[Generative AI]] ... [[Conversational AI]] ... [[ChatGPT]] | [[OpenAI]] ... [[Bing]] | [[Microsoft]] ... [[Bard]] | [[Google]] ... [[Claude]] | [[Anthropic]] ... [[Perplexity]] ... [[You]] ... [[Ernie]] | [[Baidu]]
+* [[What is Artificial Intelligence (AI)? | Artificial Intelligence (AI)]] ... [[Generative AI]] ... [[Machine Learning (ML)]] ... [[Deep Learning]] ... [[Neural Network]] ... [[Reinforcement Learning (RL)|Reinforcement]] ... [[Learning Techniques]]

@@ Line 5: / Line 5: @@
 |description=Helpful resources for your journey with artificial intelligence; videos, articles, techniques, courses, profiles, and tools
 }}
-[http://www.youtube.com/results?search_query=Variational+Autoencoders YouTube search...]
+[https://www.youtube.com/results?search_query=Variational+Autoencoders YouTube search...]
-[http://www.google.com/search?q=Variational+Autoencoders+machine+learning+ML+artificial+intelligence ...Google search]
+[https://www.google.com/search?q=Variational+Autoencoders+machine+learning+ML+artificial+intelligence ...Google search]
-* [http://www.nature.com/articles/s41587-019-0224-x#MOESM3 Deep learning enables rapid identification of potent DDR1 kinase inhibitors | A. Zhavoronkov, Y. Ivanenkov, A. Aliper, M. Veselov, V. Aladinskiy, A. Aladinskaya, V. Terentiev, D. Polykovskiy, M. Kuznetsov, A. Asadulaev, Y. Volkov, A. Zholus, R. Shayakhmetov, A. Zhebrak, L. Minaeva, B. Zagribelnyy, L. Lee, R. Soll, D. Madge, L. Xing, T. Guo & A. Aspuru-Guzik - Nature  Biotechnology]
+* [https://www.nature.com/articles/s41587-019-0224-x#MOESM3 Deep learning enables rapid identification of potent DDR1 kinase inhibitors | A. Zhavoronkov, Y. Ivanenkov, A. Aliper, M. Veselov, V. Aladinskiy, A. Aladinskaya, V. Terentiev, D. Polykovskiy, M. Kuznetsov, A. Asadulaev, Y. Volkov, A. Zholus, R. Shayakhmetov, A. Zhebrak, L. Minaeva, B. Zagribelnyy, L. Lee, R. Soll, D. Madge, L. Xing, T. Guo & A. Aspuru-Guzik - Nature  Biotechnology]
-** [http://github.com/insilicomedicine/gentrl GENTRL | GitHub]
+** [https://github.com/insilicomedicine/gentrl GENTRL | GitHub]
-** [http://www.nature.com/articles/s41587-019-0224-x#MOESM1 Supplementary Code]
+** [https://www.nature.com/articles/s41587-019-0224-x#MOESM1 Supplementary Code]
-** [http://www.cxotalk.com/episode/ai-medicine-life-sciences-drug-discovery AI in Medicine: Life Sciences and Drug Discovery - Transcript | Alex Zhavoronkov & Michael Krigsman]
+** [https://www.cxotalk.com/episode/ai-medicine-life-sciences-drug-discovery AI in Medicine: Life Sciences and Drug Discovery - Transcript | Alex Zhavoronkov & Michael Krigsman]
-** [http://singularityhub.com/2019/09/09/how-an-ai-startup-designed-a-drug-candidate-in-just-46-days/ How an AI Startup Designed a Drug Candidate in Just 46 Days | Edd Gent - SingularityHub]
+** [https://singularityhub.com/2019/09/09/how-an-ai-startup-designed-a-drug-candidate-in-just-46-days/ How an AI Startup Designed a Drug Candidate in Just 46 Days | Edd Gent - SingularityHub]
 * [[Variational Autoencoder (VAE)]]
 * [[Generative Adversarial Network (GAN)]]
@@ Line 22: / Line 22: @@
-<img src="http://github.com/insilicomedicine/GENTRL/raw/master/images/gentrl.png" alt="GENTRL" width="800" height="800">
+<img src="https://github.com/insilicomedicine/GENTRL/raw/master/images/gentrl.png" alt="GENTRL" width="800" height="800">
 <youtube>-_pHs2PLq1o</youtube>
@@ Line 31: / Line 31: @@
 <youtube>kpfB_I6QZ1w</youtube>
-http://icdn9.digitaltrends.com/image/digitaltrends/gan-chemistry-gan-timeline-original-1200x9999.jpg
+https://icdn9.digitaltrends.com/image/digitaltrends/gan-chemistry-gan-timeline-original-1200x9999.jpg