João Graça
Lisboa, Lisboa, Portugal
10 mil seguidores
+ de 500 conexões
Atividades
-
Well, that’s it, after 30+ years I’ve completed my last day in Dublin City University. I’ve been in this office for the past 11 years, and now it’ll…
Well, that’s it, after 30+ years I’ve completed my last day in Dublin City University. I’ve been in this office for the past 11 years, and now it’ll…
João Graça gostou
-
Growing up with my novelty-seeking ADHD, I often doubted if I could focus on one partner for a lifetime—or if anyone could handle me for that long!…
Growing up with my novelty-seeking ADHD, I often doubted if I could focus on one partner for a lifetime—or if anyone could handle me for that long!…
João Graça gostou
-
I’m thrilled to be part of the Scaling Up Program by Unicorn Factory Lisboa, where I have the privilege of mentoring some of the most promising…
I’m thrilled to be part of the Scaling Up Program by Unicorn Factory Lisboa, where I have the privilege of mentoring some of the most promising…
Compartilhado por João Graça
Licenças e certificados
Publicações
-
Posterior Sparsity in Dependency Grammar Induction
Journal of Machine Learning Research
A strong inductive bias is essential in unsupervised grammar induction. In this paper, we explore a particular sparsity bias in dependency grammars that encourages a small number of unique dependency types. We use part-of-speech (POS) tags to group dependencies by parent-child types and investigate sparsity-inducing penalties on the posterior distributions of parent-child POS tag pairs in the posterior regularization (PR) framework of Graça et al. (2007). In experiments with 12 different…
A strong inductive bias is essential in unsupervised grammar induction. In this paper, we explore a particular sparsity bias in dependency grammars that encourages a small number of unique dependency types. We use part-of-speech (POS) tags to group dependencies by parent-child types and investigate sparsity-inducing penalties on the posterior distributions of parent-child POS tag pairs in the posterior regularization (PR) framework of Graça et al. (2007). In experiments with 12 different languages, we achieve significant gains in directed attachment accuracy over the standard expectation maximization (EM) baseline, with an average accuracy improvement of 6.5%, outperforming EM by at least 1% for 9 out of 12 languages. Furthermore, the new method outperforms models based on standard Bayesian sparsity-inducing parameter priors with an average improvement of 5% and positive gains of at least 1% for 9 out of 12 languages. On English text in particular, we show that our approach improves performance over other state-of-the-art techniques.
Outros autores -
Controlling Complexity in Part-of-Speech Induction
Journal of Artificial Intelligence Research
We consider the problem of fully unsupervised learning of grammatical (part-of-speech) categories from unlabeled text. The standard maximum-likelihood hidden Markov model for this task performs poorly, because of its weak inductive bias and large model capacity. We address this problem by refining the model and modifying the learning objective to control its capacity via parametric and non-parametric constraints. Our approach enforces word-category association sparsity, adds morphological and…
We consider the problem of fully unsupervised learning of grammatical (part-of-speech) categories from unlabeled text. The standard maximum-likelihood hidden Markov model for this task performs poorly, because of its weak inductive bias and large model capacity. We address this problem by refining the model and modifying the learning objective to control its capacity via parametric and non-parametric constraints. Our approach enforces word-category association sparsity, adds morphological and orthographic features, and eliminates hard-to-estimate parameters for rare words. We develop an efficient learning algorithm that is not much more computationally intensive than standard training. We also provide an open-source implementation of the algorithm. Our experiments on five diverse languages (Bulgarian, Danish, English, Portuguese, Spanish) achieve significant improvements compared with previous methods for the same task.
Outros autores -
Learning Tractable Word Alignment Models with Complex Constraints
Journal of Computational Linguistics
Word-level alignment of bilingual text is a critical resource for a growing variety of tasks. Probabilistic models forward alignment present a fundamental trade-off between richness of captured constraints and correlations versus efficiency and tractability of inference. In this article, we use the Posterior Regularization framework (Graça, Ganchev, and Taskar 2007) to incorporate complex constraints into probabilistic models during learning without changing the efficiency of the underlying…
Word-level alignment of bilingual text is a critical resource for a growing variety of tasks. Probabilistic models forward alignment present a fundamental trade-off between richness of captured constraints and correlations versus efficiency and tractability of inference. In this article, we use the Posterior Regularization framework (Graça, Ganchev, and Taskar 2007) to incorporate complex constraints into probabilistic models during learning without changing the efficiency of the underlying model. We focus on the simple and tractable hidden Markov model, and present an efficient learning algorithm for incorporating approximate bijectivity and symmetry constraints. Models estimated with these constraints produce a significant boost in performance as measured by both precision and recall of manually annotated alignments for six language pairs. We also report experiments on two different tasks where word alignments are required: phrase based machine translation and syntax transfer, and show promising improvements over standard methods.
Outros autores -
Posterior Regularization in Latent Variable Models
Journal of Machine Learning Research
We present posterior regularization, a probabilistic framework for structured, weakly supervised learning. Our framework efficiently incorporates indirect supervision via constraints on posterior distributions of probabilistic models with latent variables. Posterior regularization separates model complexity from the complexity of structural constraints it is desired to satisfy. By directly imposing decomposable regularization on the posterior moments of latent variables during learning, we…
We present posterior regularization, a probabilistic framework for structured, weakly supervised learning. Our framework efficiently incorporates indirect supervision via constraints on posterior distributions of probabilistic models with latent variables. Posterior regularization separates model complexity from the complexity of structural constraints it is desired to satisfy. By directly imposing decomposable regularization on the posterior moments of latent variables during learning, we retain the computational efficiency of the unconstrained model while ensuring desired constraints
hold in expectation. We present an efficient algorithm for learning with posterior regularization and illustrate its versatility on a diverse set of structural constraints such as bijectivity, symmetry and group sparsity in several large scale experiments, including multi-view learning, cross-lingual dependency grammar induction, unsupervised part-of-speech induction, and bitext word alignment.Outros autores
Organizações
-
LXMLS - Lisbon Machine Learning Summer School
Main Organizer
One week Summer School about core Machine Learning and Natural Language Processing methods and their applications. www.lxmls.it.pt
Mais atividade de João
-
Three of the five toughest University programs to get admission to in Portugal are aerospace engineering degrees. Does this make any sense? Are…
Three of the five toughest University programs to get admission to in Portugal are aerospace engineering degrees. Does this make any sense? Are…
João Graça gostou
-
Last week we signed our first $1M contract! 🍾 A huge milestone for the company, which reminded me of our humble beginnings... About 9 years ago, I…
Last week we signed our first $1M contract! 🍾 A huge milestone for the company, which reminded me of our humble beginnings... About 9 years ago, I…
João Graça gostou
-
Few people know that there is a common thread that runs through how Apple, Google, and Amazon built their legendary teams 🧶 All of their founders…
Few people know that there is a common thread that runs through how Apple, Google, and Amazon built their legendary teams 🧶 All of their founders…
João Graça gostou
-
How can we create smaller, more efficient LLMs from larger ones? 🤔 NVIDIA combined structured weight pruning with knowledge distillation to reduce…
How can we create smaller, more efficient LLMs from larger ones? 🤔 NVIDIA combined structured weight pruning with knowledge distillation to reduce…
João Graça gostou
-
💥Awesome news alert💥 Par Equity portfolio company Aveni has secured £11m in one of Scotland's largest Series A raises this year!…
💥Awesome news alert💥 Par Equity portfolio company Aveni has secured £11m in one of Scotland's largest Series A raises this year!…
João Graça gostou
-
I think that neither Trump nor Harris will destroy America if elected president. Call me crazy, but I think that Trump is not a fascist and Harris…
I think that neither Trump nor Harris will destroy America if elected president. Call me crazy, but I think that Trump is not a fascist and Harris…
João Graça gostou
-
Sword Health just announced that the Sword Foundation will partner with non-profit organizations to provide their solutions at no cost and fund…
Sword Health just announced that the Sword Foundation will partner with non-profit organizations to provide their solutions at no cost and fund…
João Graça gostou
-
The WMT24 conference has released a preliminary report, 📊 offering an early look 🔎 at the rankings of general #MT systems and #LLMs. Findings based…
The WMT24 conference has released a preliminary report, 📊 offering an early look 🔎 at the rankings of general #MT systems and #LLMs. Findings based…
João Graça gostou
Outros perfis semelhantes
Outras pessoas chamadas João Graça em Portugal
-
João Graça
-
João Graça
Head of B2C One App at Santander Portugal | Lead Product Owner
-
João Graça
𝗧𝗮𝗹𝗲𝗻𝘁 𝗦𝗽𝗲𝗰𝗶𝗮𝗹𝗶𝘀𝘁 | 𝗫𝗽𝗮𝗻𝗱 𝗜𝗧
-
João Graça
Tech Recruiter
-
João Graça
Mais 200 pessoas chamadas João Graça fazem parte do LinkedIn em Portugal
Veja mais pessoas chamadas João Graça