scholarly journals Revision in Continuous Space: Unsupervised Text Style Transfer without Adversarial Learning

2020 ◽  
Vol 34 (05) ◽  
pp. 8376-8383
Author(s):  
Dayiheng Liu ◽  
Jie Fu ◽  
Yidan Zhang ◽  
Chris Pal ◽  
Jiancheng Lv

Typical methods for unsupervised text style transfer often rely on two key ingredients: 1) seeking the explicit disentanglement of the content and the attributes, and 2) troublesome adversarial learning. In this paper, we show that neither of these components is indispensable. We propose a new framework that utilizes the gradients to revise the sentence in a continuous space during inference to achieve text style transfer. Our method consists of three key components: a variational auto-encoder (VAE), some attribute predictors (one for each attribute), and a content predictor. The VAE and the two types of predictors enable us to perform gradient-based optimization in the continuous space, which is mapped from sentences in a discrete space, to find the representation of a target sentence with the desired attributes and preserved content. Moreover, the proposed method naturally has the ability to simultaneously manipulate multiple fine-grained attributes, such as sentence length and the presence of specific words, when performing text style transfer tasks. Compared with previous adversarial learning based methods, the proposed method is more interpretable, controllable and easier to train. Extensive experimental studies on three popular text style transfer tasks show that the proposed method significantly outperforms five state-of-the-art methods.

Author(s):  
Yaniv Aspis ◽  
Krysia Broda ◽  
Alessandra Russo ◽  
Jorge Lobo

We introduce a novel approach for the computation of stable and supported models of normal logic programs in continuous vector spaces by a gradient-based search method. Specifically, the application of the immediate consequence operator of a program reduct can be computed in a vector space. To do this, Herbrand interpretations of a propositional program are embedded as 0-1 vectors in $\mathbb{R}^N$ and program reducts are represented as matrices in $\mathbb{R}^{N \times N}$. Using these representations we prove that the underlying semantics of a normal logic program is captured through matrix multiplication and a differentiable operation. As supported and stable models of a normal logic program can now be seen as fixed points in a continuous space, non-monotonic deduction can be performed using an optimisation process such as Newton's method. We report the results of several experiments using synthetically generated programs that demonstrate the feasibility of the approach and highlight how different parameter values can affect the behaviour of the system.


2021 ◽  
Author(s):  
Yiwei Lyu ◽  
Paul Pu Liang ◽  
Hai Pham ◽  
Eduard Hovy ◽  
Barnabás Póczos ◽  
...  
Keyword(s):  

Author(s):  
Asım Balbay ◽  
Engin Avci ◽  
Ömer Şahin ◽  
Resul Coteli

Abstract Artificial neural networks (ANNs) have been widely used in modeling of various systems. Training of ANNs is commonly performed by backpropagation based on a gradient-based learning rule. However, it is well-known that such learning rule has several shortcomings such as slow convergence and training failures. This paper proposes a modeling technique based on Extreme Learning Machine (ELM) eliminating disadvantages of backpropagation based on a gradient-based learning rule for the drying of bittim (pistacia terebinthus). The samples for ELM based model are obtained by experimental studies. In experimental studies, the sample mass loss rate as a function time was investigated in different air velocities (0.5 and 1 m/s) and air temperatures (40, 60 and 80°C) in a designed dryer system. The obtained samples from experiments are used for training and testing of ELM. Further, some parameters of ELM such as type of activation function and the number of hidden neurons are set to obtain the best possible modelling results. The obtained prediction results show that ELM algorithm with tangent sigmoid activation function and 20 hidden neurons is appeared to be most optimal topology since maximum R2 and minimum rms (0.0500) and cov (0.2256) values are obtained. Thus, it is concluded that ELM can be used as an effective modelling tool in the drying of bittim (pistacia terebinthus) in fixed bed dryer system.


Author(s):  
Kikuo Fujita ◽  
Tomoki Ushiro ◽  
Noriyasu Hirokawa

This paper proposes a new design optimization framework by integrating evolutionary search and cumulative function approximation. While evolutionary algorithms are robust even under multi-peaks, rugged natures, etc., their computational cost is inferior to ordinary schemes such as gradient-based methods. While response surface techniques such as quadratic approximation can save computational cost for complicated design problems, the fidelity of solution is affected by density of samples. The new framework simultaneously performs evolutionary search and constructs response surfaces. That is, in its early phase the search is performed over roughly but globally approximated surfaces with the relatively small number of samples, and in its later phase the search is performed intensively around promising regions, which are revealed in the preceded phases, over response surfaces enhanced with additional samples. This framework is expected to be able to robustly find the optimal solution with less sampling. An optimization algorithm is implemented by combining a real-coded genetic algorithm and a Voronoi diagram based cumulative approximation, and it is applied to some numerical examples for discussing its potential and promises.


2018 ◽  
Vol 251 ◽  
pp. 02002 ◽  
Author(s):  
Sergey Parfenov ◽  
Anatoly Alekseytsev ◽  
Yuriy Vinokurov

Describes the theoretical preconditions of using mechanical properties of fine-grained concrete in the design of concrete structures subject to risks and the general safety. A technique for experimental studies and data on the fine-grained concrete deformative properties at different ages and different loading levels are presented. The regularities of the modulus of elasticity change from strength, type and age of concrete are revealed. Full diagrams of deformation of concrete are constructed. The results obtained can be used in the design of the fine-grained concrete structures in buildings having risks occurrence socioeconomic losses and able to resist of emergency actions.


Electronics ◽  
2021 ◽  
Vol 11 (1) ◽  
pp. 37
Author(s):  
Hai-Tao Yu ◽  
Degen Huang ◽  
Fuji Ren ◽  
Lishuang Li

Learning-to-rank has been intensively studied and has shown significantly increasing values in a wide range of domains, such as web search, recommender systems, dialogue systems, machine translation, and even computational biology, to name a few. In light of recent advances in neural networks, there has been a strong and continuing interest in exploring how to deploy popular techniques, such as reinforcement learning and adversarial learning, to solve ranking problems. However, armed with the aforesaid popular techniques, most studies tend to show how effective a new method is. A comprehensive comparison between techniques and an in-depth analysis of their deficiencies are somehow overlooked. This paper is motivated by the observation that recent ranking methods based on either reinforcement learning or adversarial learning boil down to policy-gradient-based optimization. Based on the widely used benchmark collections with complete information (where relevance labels are known for all items), such as MSLRWEB30K and Yahoo-Set1, we thoroughly investigate the extent to which policy-gradient-based ranking methods are effective. On one hand, we analytically identify the pitfalls of policy-gradient-based ranking. On the other hand, we experimentally compare a wide range of representative methods. The experimental results echo our analysis and show that policy-gradient-based ranking methods are, by a large margin, inferior to many conventional ranking methods. Regardless of whether we use reinforcement learning or adversarial learning, the failures are largely attributable to the gradient estimation based on sampled rankings, which significantly diverge from ideal rankings. In particular, the larger the number of documents per query and the more fine-grained the ground-truth labels, the greater the impact policy-gradient-based ranking suffers. Careful examination of this weakness is highly recommended for developing enhanced methods based on policy gradient.


1992 ◽  
Vol 6 ◽  
pp. 38-38
Author(s):  
Thomas W. Broadhead ◽  
Steven G. Driese

Small skeletal sediment particles, by virtue of their size, experience physical sedimentological conditions different from those affecting large (i.e., > 4 mm) particles. In shallow, high-energy marine carbonate environments with an overall fine-grained (< 4 mm) sediment composition, grain-to-grain impacts are probably of insufficient magnitude to overcome the cushioning effect of water and cause abrasive wear. Very large skeletal grains within such a setting will promote physical degradation, but biological activity (microboring, bioerosion) is more effective in modifying particle surfaces and diminishing grain size. Only in eolian environments are collisions between small grains effective in causing abrasion. Quartz sand, a particularly important abrasive agent in the destruction of large skeletal parts under even moderate energy regimes, also appears to be appreciably important for the degradation of small skeletons only under eolian conditions.Conodonts are small (0.2–2.0 mm), phosphatic teeth, which may have belonged to an extinct (Cambrian-Triassic) group of jawless fish. They occur principally in marine and marginal marine sedimentary rocks; most specimens probably were deposited as discrete particles or as components of fecal masses. Because the extent and nature of degradation of a skeleton reflects its taphonomic history, evaluation of the susceptibility of conodonts to physical abrasion is important. Bioerosion, probably the principal cause of degradation of calcareous skeletons, was probably insignificant in alteration of conodonts. Their mineralogical composition made conodonts unlikely substrates for endoliths, and they were insufficiently abundant to serve as a recyclable source of phosphate for larger metazoans. Chemical breakdown of skeletal apatite would not have occurred, although atmospheric exposure may have promoted degradation of organic laminae, causing exfoliation.Experimental studies of abrasion (elements of Palmatolepis sp., Polygnathus sp. - U. Devonian, Iowa) plus petrographic examination of conodont bearing marine, marginal marine, and eolian facies of the Morgan Formation (M. Pennsylvanian, Utah and Colorado) suggests that abrasion of conodonts is insignificant in aqueous environments, and likely to occur to a major degree only under eolian conditions. High-energy (25.6 cm/sec) conditions were simulated in a tumbler containing quartz sand (4Φ − 2Φ), artificial sea water and moderate conodont abundances (50 specimens/kg). Abrasion produced under these artificially extreme conditions (e.g., continuous motion for up to 20 km travel distance) was at a scale detectable only with SEM. “Dry” tumbling experiments were ineffective, as noted by other authors simulating conditions for abrasion of mineral grains. Eolian conditions were simulated in a glass “wind tube” (see Kuenen, 1960, Journal of Geology) using a similar range of quartz sand sizes. Air velocity of approximately 100 cm/sec produced subtle visually detectable abrasion following 5 hours of exposure/transport in 2Φ quartz sand (1667 specimens/kg), but exhibited no visual evidence of abrasion under similar conditions with 3Φ sand.The Morgan Formation is a mixed carbonate/siliciclastic sequence that reflects repeated episodes of submergence and exposure of a carbonate shelf and laterally adjacent siliciclastic eolian “sand sea.” It represents a virtually ideal ancient example for the interpretation of the effects of natural sedimentary processes because of its well-documented environmental heterogeneity and contains both abraded and unabraded conodonts. Conodonts sampled from high-energy subtidal carbonate facies of the Morgan are unabraded, irrespective of the presence of quartz sand; platform and delicate ramiform elements are all represented. Conodonts sampled from mixed siliciclastic/carbonate eolian facies are extremely abraded; only the most durable Pa elements of Adetognathus sp. and Idiognathodus sp. are preserved, although fine surface details and denticles are worn and anterior blades are missing. Laboratory simulations confirm the interpretation that marine processes are generally insufficient to physically abrade small phosphatic skeletal grains and that highly abraded Morgan conodonts experienced extremely long residence times in the eolian environment where they were transported, repeatedly buried, and exposed to bombardment and abrasion by quartz sand.Supported by NSF EAR9004300


2020 ◽  
Vol 10 (18) ◽  
pp. 6196
Author(s):  
Chunhua Wu ◽  
Xiaolong Chen ◽  
Xingbiao Li

Currently, most text style transfer methods encode the text into a style-independent latent representation and decode it into new sentences with the target style. Due to the limitation of the latent representation, previous works can hardly get satisfactory target style sentence especially in terms of semantic remaining of the original sentence. We propose a “Mask and Generation” structure, which can obtain an explicit representation of the content of original sentence and generate the target sentence with a transformer. This explicit representation is a masked text that masks the words with the strong style attribute in the sentence. Therefore, it can preserve most of the semantic meaning of the original sentence. In addition, as it is the input of the generator, it also simplified this process compared to the current work who generate the target sentence from scratch. As the explicit representation is readable and the model has better interpretability, we can clearly know which words changed and why the words changed. We evaluate our model on two review datasets with quantitative, qualitative, and human evaluations. The experimental results show that our model generally outperform other methods in terms of transfer accuracy and content preservation.


2019 ◽  
Vol 33 (34) ◽  
pp. 1950427
Author(s):  
Liang Zhao ◽  
Hongan Ma ◽  
Chao Fang ◽  
Luyao Ding ◽  
Baomin Liu ◽  
...  

According to the mineral composition [Formula: see text] of jadeite, [Formula: see text] and [Formula: see text] with a molar ratio of 1:1 were selected as raw materials. One sample was undoped, and five samples were mixed with 0.4 wt.% [Formula: see text], MnO, CoO, [Formula: see text] and [Formula: see text], respectively. The experimental studies were executed under the synthetic condition of 5 GPa pressure and [Formula: see text] temperature using China-type large volume cubic high-pressure apparatus (CHPA) (SPD-6X1200). After the experiment, compositions of the synthetic jadeites were characterized by X-ray diffraction, microstructures characterized by field emission scanning electron microscopy (FESEM) and the high-resolution transmission electron microscopy (HRTEM), molecular vibration types investigated by Raman spectra. The color-causing ion of the synthetic jadeites was investigated by UV–Vis Spectra. The results show that the synthetic jadeites have excellent crystallinity and fine-grained texture and similar structural behavior with natural jadeite. UV–Vis Spectra indicated that different colors of samples have different ion absorption peaks, thus showing different colors. This experiment explored the geological conditions of synthetic jadeite by means of high pressure and high temperature (HPHT) to give a favorable experimental basis for synthetic jadeite and analyze the coloration mechanism of jadeite by way of ion-doping.


Sign in / Sign up

Export Citation Format

Share Document