Zobrazeno 1 - 10
of 233
pro vyhledávání: '"Kervadec A"'
Autor:
Cheng, Emily, Doimo, Diego, Kervadec, Corentin, Macocco, Iuri, Yu, Jade, Laio, Alessandro, Baroni, Marco
A language model (LM) is a mapping from a linguistic context to an output token. However, much remains to be known about this mapping, including how its geometric properties relate to its function. We take a high-level geometric approach to its analy
Externí odkaz:
http://arxiv.org/abs/2405.15471
Autor:
Breznik, Eva, Kervadec, Hoel, Malmberg, Filip, Kullberg, Joel, Ahlström, Håkan, de Bruijne, Marleen, Strand, Robin
This paper investigates the combination of intensity-based distance maps with boundary loss for point-supervised semantic segmentation. By design the boundary loss imposes a stronger penalty on the false positives the farther away from the object the
Externí odkaz:
http://arxiv.org/abs/2311.03537
Language model prompt optimization research has shown that semantically and grammatically well-formed manually crafted prompts are routinely outperformed by automatically generated token sequences with no apparent meaning or syntactic structure, incl
Externí odkaz:
http://arxiv.org/abs/2310.15829
For a language model (LM) to faithfully model human language, it must compress vast, potentially infinite information into relatively few dimensions. We propose analyzing compression in (pre-trained) LMs from two points of view: geometric and informa
Externí odkaz:
http://arxiv.org/abs/2310.13620
Autor:
Kervadec, Hoel, de Bruijne, Marleen
In the past few years, in the context of fully-supervised semantic segmentation, several losses -- such as cross-entropy and dice -- have emerged as de facto standards to supervise neural networks. The Dice loss is an interesting case, as it comes fr
Externí odkaz:
http://arxiv.org/abs/2304.04319
Autor:
Cristiana Dondi, Georg Vogler, Anjali Gupta, Stanley M. Walls, Anaïs Kervadec, James Marchant, Michaela R. Romero, Soda Diop, Jason Goode, John B. Thomas, Alex R. Colas, Rolf Bodmer, Marc Montminy, Karen Ocorr
Publikováno v:
Cell Reports, Vol 43, Iss 8, Pp 114549- (2024)
Summary: CREB-regulated transcription co-activator (CRTC) is activated by Calcineurin (CaN) to regulate gluconeogenic genes. CaN also has roles in cardiac hypertrophy. Here, we explore a cardiac-autonomous role for CRTC in cardiac hypertrophy. In Dro
Externí odkaz:
https://doaj.org/article/c996134180b94600beaae178614fec84
As in many tasks combining vision and language, both modalities play a crucial role in Visual Question Answering (VQA). To properly solve the task, a given model should both understand the content of the proposed image and the nature of the question.
Externí odkaz:
http://arxiv.org/abs/2202.06858
Autor:
Dondi, Cristiana, Vogler, Georg, Gupta, Anjali, Walls, Stanley M., Kervadec, Anaïs, Marchant, James, Romero, Michaela R., Diop, Soda, Goode, Jason, Thomas, John B., Colas, Alex R., Bodmer, Rolf, Montminy, Marc, Ocorr, Karen
Publikováno v:
In Cell Reports 27 August 2024 43(8)
Domain adaptation (DA) has drawn high interest for its capacity to adapt a model trained on labeled source data to perform well on unlabeled or weakly labeled target data from a different domain. Most common DA techniques require concurrent access to
Externí odkaz:
http://arxiv.org/abs/2108.03152
Methods for Visual Question Anwering (VQA) are notorious for leveraging dataset biases rather than performing reasoning, hindering generalization. It has been recently shown that better reasoning patterns emerge in attention layers of a state-of-the-
Externí odkaz:
http://arxiv.org/abs/2106.05597