RedCircle VERIFY_TOKEN_3bf9dea5-8505-4c5f-9864-8ba5090c504f -- Rendered At Sat, 14 Nov 2026 04:57:00 +0000 Pap Series https://redcircle.com/shows/pap-series en Sanket Thakur 2021 Sanket Thakur This is a series which I recently started to talk to people about their current work and discuss any one of their favorite papers. 3bf9dea5-8505-4c5f-9864-8ba5090c504f This is a series which I recently started to talk to people about their current work and discuss any one of their favorite papers.

]]> episodic no Sanket Thakur sanketkumar1996@gmail.com https://feeds.redcircle.com/3bf9dea5-8505-4c5f-9864-8ba5090c504f false full Bayesian for large image data ft. Siddharth Swaroop Bayesian for large image data ft. Siddharth Swaroop 5 1 Sanket Thakur In this episode, I talk to Siddharth Swaroop of Cambridge University about his paper 'Practical Deep Learning with Bayesian Principles' on how bayesian methods can be exploited for large image datasets and compare with existing deep learning models.

]]> <p>In this episode, I talk to Siddharth Swaroop of Cambridge University about his paper &#39;Practical Deep Learning with Bayesian Principles&#39; on how bayesian methods can be exploited for large image datasets and compare with existing deep learning models. </p> 891999be-d1b4-4153-be44-587b6e45a45a https://siddharthswaroop.github.io Fri, 04 Feb 2022 21:47:04 +0000 970 false full Large Language Models ft. Haau-Sing Li Large Language Models ft. Haau-Sing Li 4 1 Sanket Thakur In this episode, I talk to Haau-Sing Li, a Ph.D. student at TU Darmstadt working on multi modal learning focusing on code generation. We talk about the paper 'large language models on code' and related discussion about codex and GPT-3.

Haau-Sing Li : https://www.linkedin.com/in/haau-sing-li-152177142/?originalSubdomain=de

Timestamps :

  • (1:03) - Podcast starts
  • (3:48) - Paper discussion
  • (14:24) - Is it easier to deal with NL than code data ?
  • (22:48) - Can AI replace programmers job ?
  • (32:38) - Music

Haau-Sing's Music band on Spotify :

]]> <p>In this episode, I talk to Haau-Sing Li, a Ph.D. student at TU Darmstadt working on multi modal learning focusing on code generation. We talk about the paper &#39;large language models on code&#39; and related discussion about codex and GPT-3.</p><p>Haau-Sing Li : <a href="https://www.linkedin.com/in/haau-sing-li-152177142/?originalSubdomain=de" rel="nofollow">https://www.linkedin.com/in/haau-sing-li-152177142/?originalSubdomain=de </a></p><p>Timestamps :</p><ul><li>(1:03) - Podcast starts</li><li>(3:48) - Paper discussion</li><li>(14:24) - Is it easier to deal with NL than code data ?</li><li>(22:48) - Can AI replace programmers job ?</li><li>(32:38) - Music</li></ul><p>Haau-Sing&#39;s Music band on <a href="https://open.spotify.com/album/2XUTS3fjhigrGNBi1jR4Nz?si=f-Ru8xNrR9mHc0z_grrNXA" rel="nofollow">Spotify</a> :</p> a708efe3-72ea-4d88-afca-2c2556e5dd38 https://rss.com/podcasts/papseries Sun, 21 Nov 2021 22:15:36 +0000 2134 false full AI safety ft. Marius Hobbhan AI safety ft. Marius Hobbhan 3 1 Sanket Thakur In this episode, I talk to Marius Hobbhan about the paper 'Concrete problems in AI safety' and related aspects of it. We also happen to talk about 'Effective altruism ' and how one can join a local chapter. <br><br>

<br>

Timestamps : <br>

  • (01:04) : Podcast starts <br>
  • (5:24) : Paper discussion : Concrete problems in AI safety <br>
  • (21:34) : Is AI smart enough ? <br>
  • (23:27): Effective altruism <br>
]]> <p>In this episode, I talk to Marius Hobbhan about the paper &#39;Concrete problems in AI safety&#39; and related aspects of it. We also happen to talk about &#39;Effective altruism &#39; and how one can join a local chapter. &lt;br&gt;&lt;br&gt;</p><ul><li>Marius website : <a href="https://www.mariushobbhahn.com" rel="nofollow">https://www.mariushobbhahn.com </a>&lt;br&gt;</li><li>Effective altruism : <a href="https://www.effectivealtruism.org" rel="nofollow">https://www.effectivealtruism.org</a> &lt;br&gt;</li><li>80000 hours : <a href="https://80000hours.org" rel="nofollow">https://80000hours.org</a> &lt;br&gt;</li></ul><p>&lt;br&gt;</p><p>Timestamps : &lt;br&gt;</p><ul><li>(01:04) : Podcast starts &lt;br&gt;</li><li>(5:24) : Paper discussion : Concrete problems in AI safety &lt;br&gt;</li><li>(21:34) : Is AI smart enough ? &lt;br&gt;</li><li>(23:27): Effective altruism &lt;br&gt;</li></ul> 5de303c2-f665-4c17-a2f0-fedf94a8f1c4 https://rss.com/podcasts/papseries Sun, 07 Nov 2021 18:38:46 +0000 1793 false full Uncertainty ft. Gianluca Scarpellini Uncertainty ft. Gianluca Scarpellini 2 1 Sanket Thakur In the episode, Gianluca talks about uncertainty in deep learning, why is it important and how it can potentially be implemented. <br>

Besides, he also discuss about his work outside of academics. <br>

Yarin gal uncertainty demo / simulator : <br>

Gianluca profile : https://github.com/gianscarpe <br>

Plug in as the podcast starts very technical and gets quite funny by the end.

]]> <p>In the episode, Gianluca talks about uncertainty in deep learning, why is it important and how it can potentially be implemented. &lt;br&gt;</p><p>Besides, he also discuss about his work outside of academics. &lt;br&gt;</p><p>Yarin gal uncertainty <a href="https://htmlpreview.github.io/?https%3A%2F%2Fgithub.com%2Fyaringal%2FHeteroscedasticDropoutUncertainty%2Fblob%2Fmaster%2Fdemos%2Fheteroscedastic_dropout_reg.html=" rel="nofollow">demo / simulator</a> : &lt;br&gt;</p><p>Gianluca profile : <a href="https://github.com/gianscarpe" rel="nofollow">https://github.com/gianscarpe</a> &lt;br&gt;</p><p>Plug in as the podcast starts very technical and gets quite funny by the end.</p> 66b96164-b847-45bf-8b02-e6ae001ce0ee https://rss.com/podcasts/papseries Sun, 17 Oct 2021 15:49:35 +0000 1451 false full Fairness explained ft. Seamus Ryan Fairness explained ft. Seamus Ryan 1 1 Sanket Thakur In this first episode of Pap series, we talk to Seamus Ryan about fairness in machine learning and how it can be used to solve fundamental problems. In this first episode of Pap series, I talk to Seamus Ryan about fairness in machine learning and how it can be used to solve fundamental problems.

]]> <p>In this first episode of Pap series, I talk to Seamus Ryan about fairness in machine learning and how it can be used to solve fundamental problems.</p> 28acd7d7-90d5-2ff3-f9bb-6b61cb313a8d https://rss.com/podcasts/papseries/249057 Tue, 03 Aug 2021 14:20:47 +0000 974 false