2,162 research outputs found
Text classification by untrained sentence embeddings
No full textRecurrent Neural Networks (RNNs) represent a natural paradigm for modeling sequential data like text written in natural language. In fact, RNNs and their variations have long been the architecture of choice in many applications, however in practice they require the use of labored architectures (such as gating mechanisms) and computationally heavy training processes. In this paper we address the question of whether it is possible to generate sentence embeddings via completely untrained recurrent dynamics, on top of which to apply a simple learning algorithm for text classification. This would allow to obtain extremely efficient models in terms of training time. Our work investigates the extent to which this approach can be used, by analyzing the results on different tasks. Finally, we show that, within certain limits, it is possible to build extremely efficient models for text classification that remain competitive in accuracy with reference models in the state-of-the-art
Una Memoria per costruire il futuro, conferenza di Moni Ovadia
No full textConferenza organizzata in collaborazione con Associazione Culturale Balamòs e Teatro de Micheli di Copparo, presso Aula Drigo, Dipartimento di Scienze Storich
Robustness Envelopes for Temporal Plans
No full textTo achieve practical execution, planners must produce temporal plans with some degree of run-time adaptability. Such plans can be expressed as Simple Temporal Networks (STN), that constrain the timing of action activations, and implicitly represent the space of choices for the plan executor.
A first problem is to verify that all the executor choices allowed by the STN plan will be successful, i.e. the plan is valid. An even more important problem is to assess the effect of discrepancies between the model used for planning and the execution environment.
We propose an approach to compute the "robustness envelope" (i.e., alternative action durations or resource consumption rates) of a given STN plan, for which the plan remains valid. Plans can have boolean and numeric variables as well as discrete and continuous change. We leverage Satisfiability Modulo Theories (SMT) to make the approach formal and practical
Question Classification with Untrained Recurrent Embeddings
No full textRecurrent Neural Networks (RNNs) are at the foundation of many state-of-the-art results in text classification. However, to be effective in practical applications, they often require the use of sophisticated architectures and training techniques, such as gating mechanisms and pre-training by autoencoders or language modeling, with typically high computational cost. In this work, we show that such techniques could actually be not always necessary. In fact, our experimental results on a Question Classification task indicate that using state-of-the-art Reservoir Computing approaches for RNN design, it is possible to achieve competitive or comparable accuracy with a considerable advantage in terms of required training times
Allometric scaling of mortality rates with body mass in abalones
No full textThe existence of an allometric relationship between mortality rates and body mass has been theorized and extensively documented across taxa. Within species, however, the allometry between mortality rates and body mass has received substantially less attention and the consistency of such scaling patterns at the intra-specific level is controversial. We reviewed 73 experimental studies to examine the relationship between mortality rates and body size among seven species of abalone (Haliotis spp.), a marine herbivorous mollusk. Both in the field and in the laboratory, log-transformed mortality rates were negatively correlated with log-transformed individual body mass for all species considered, with allometric exponents remarkably similar among species. This regular pattern confirms previous findings that juvenile abalones suffer higher mortality rates than adult individuals. Field mortality rates were higher overall than those measured in the laboratory, and the relationship between mortality and body mass tended to be steeper in field than in laboratory conditions for all species considered. These results suggest that in the natural environment, additional mortality factors, especially linked to predation, could significantly contribute to mortality, particularly at small body sizes. On the other hand, the consistent allometry of mortality rates versus body mass in laboratory conditions suggests that other sources of mortality, beside predation, are size-dependent in abalone
sj-docx-1-isp-10.1177_00207640231194478 – Supplemental material for Microaggression toward LGBTIQ people and implications for mental health: A systematic review
No full textSupplemental material, sj-docx-1-isp-10.1177_00207640231194478 for Microaggression toward LGBTIQ people and implications for mental health: A systematic review by Mattia Marchi, Antonio Travascio, Daniele Uberti, Edoardo De Micheli, Fabio Quartaroli, Giuseppe Laquatra, Pietro Grenzi, Luca Pingani, Silvia Ferrari, Andrea Fiorillo, Manlio Converti, Federica Pinna, Francesco Amaddeo, Antonio Ventriglio, Massimo Mirandola and Gian M Galeazzi in International Journal of Social Psychiatry</p
Efficient anytime computation and execution of decoupled robustness envelopes for temporal plans
No full text[Abstract unavailable
Perspettiva ridotta a perfezione: Glimpses of Daniele Barbaro’s Perspective Theory
No full textThis contribution is intended to identify some textual elements, apparently
secondary in Daniele Barbaro’s treatise on perspective, which either foreshadow
unprecedented developments in the discipline of representation or have
constituted complex critical nodes in the field of perspective. The first of these
is introduced in part V: anamorphosis (though never so called by the author, since
the term was not yet in use), suggesting a quick method to deform any flat image
by means of shadows. Finally, the author mentions two other ‘eccentric’ elements
of interest for the future developments of perspective that originated in Daniele
Barbaro’s text: an optical-projective device first introduced by Giovanni Battista
Vimercato, later developed by Jean François Niceron in his Thaumaturgus
opticus (1646), and the camera obscura
INTRUSION AND PRESENCE OF THE AUTHOR IN SAMUEL BECKETT’S “THE UNNAMABLE” AND B. S. JOHNSON’S “ALBERT ANGELO”
No full textThis article explores the intricate relationship between B.S. Johnson’s novel “Albert Angelo” and Beckett’s “The Unnamable”, both dealing with the issue of the possible presence of the author in his own text. The point of departure for such comparison is Johnson’s incorporation, as an epigraph, of a passage taken from Beckett’s novel. Such passage, intended rather literally by the British author, is employed as a justification not only for the central device at the heart of his novel, but also in support of a larger aesthetic project which will characterise a great part of his oeuvre – which famously stresses the importance of ‘truth’, as opposed to fabulation, and the necessity of the author’s direct presence in his texts. This contribution, in particular, tries to reconstruct the history of Johnson’s involvement with Beckett’s work, demonstrating how Johnson has in fact distorted the master’s message – perhaps intentionally – in order to produce a rather different model of literature, despite moving from very similar premises
- …
