BiFDANet: Unsupervised Bidirectional Domain Adaptation for Semantic Segmentation of Remote Sensing Images

When segmenting massive amounts of remote sensing images collected from different satellites or geographic locations (cities), the pre-trained deep learning models cannot always output satisfactory predictions. To deal with this issue, domain adaptation has been widely utilized to enhance the generalization abilities of the segmentation models. Most of the existing domain adaptation methods, which based on image-to-image translation, firstly transfer the source images to the pseudo-target images, adapt the classifier from the source domain to the target domain. However, these unidirectional methods suffer from the following two limitations: (1) they do not consider the inverse procedure and they cannot fully take advantage of the information from the other domain, which is also beneficial, as confirmed by our experiments; (2) these methods may fail in the cases where transferring the source images to the pseudo-target images is difficult. In this paper, in order to solve these problems, we propose a novel framework BiFDANet for unsupervised bidirectional domain adaptation in the semantic segmentation of remote sensing images. It optimizes the segmentation models in two opposite directions. In the source-to-target direction, BiFDANet learns to transfer the source images to the pseudo-target images and adapts the classifier to the target domain. In the opposite direction, BiFDANet transfers the target images to the pseudo-source images and optimizes the source classifier. At test stage, we make the best of the source classifier and the target classifier, which complement each other with a simple linear combination method, further improving the performance of our BiFDANet. Furthermore, we propose a new bidirectional semantic consistency loss for our BiFDANet to maintain the semantic consistency during the bidirectional image-to-image translation process. The experiments on two datasets including satellite images and aerial images demonstrate the superiority of our method against existing unidirectional methods.

Semantic and Syntactic Processing During Comprehension: ERP Evidence From Chinese QING Structure

Previous studies used BA and BEI structures as stimuli to infer that syntax-first models seemed not applicable in Chinese. However, there were inconsistent results of both within same structures and between different structures. Since sentence structures of stimuli were non-canonical as well as lacking wide representativeness in Chinese, we examined the processing mechanism of a more representative structure in Chinese, QING (QING + NP1 + V + NP2) structure in the current study. Four conditions, including correct sentences (CORRECT), semantic-violated sentences (SEMANTIC), syntactic-violated sentences (SYNTACTIC), and combined violated sentences (COMBINED), were composed by manipulating the V between NP1 and NP2. Results with respect to three types of violation were as follows. In the initial phrase (100–300 ms), there existed an interaction between SEMANTIC consistency and the SYNTACTIC category. In the intermediate phrase (300–500 ms), the interaction continued with similar negative waves evoked by three types of violated sentences. In the final phrase (500–700 ms), both SYNTACTIC or COMBINED evoked obvious negative waves. The current research of Qing structure provided new evidence for the processing mechanism of Chinese sentence patterns. Specifically, we found that the interactive model rather than the syntax-first model may apply to the processing of this specific structure of Chinese sentences and compared the results with those reported in previous studies that examined other types of sentence structures.

A Multi-RNN Research Topic Prediction Model Based on Spatial Attention and Semantic Consistency-Based Scientific Influence Modeling

Computer science discipline includes many research fields, which mutually influence and promote each other’s development. This poses two great challenges of predicting the research topics of each research field. One is how to model fine-grained topic representation of a research field. The other is how to model research topic of different fields and keep the semantic consistency of research topics when learning the scientific influence context from other related fields. Unfortunately, the existing research topic prediction approaches cannot handle these two challenges. To solve these problems, we employ multiple different Recurrent Neural Network chains which model research topics of different fields and propose a research topic prediction model based on spatial attention and semantic consistency-based scientific influence modeling. Spatial attention is employed in field topic representation which can selectively extract the attributes from the field topics to distinguish the importance of field topic attributes. Semantic consistency-based scientific influence modeling maps research topics of different fields to a unified semantic space to obtain the scientific influence context of other related fields. Extensive experiment results on five related research fields in the computer science (CS) discipline show that the proposed model is superior to the most advanced methods and achieves good topic prediction performance.

Recognition Memory for Pictorial Events: Fusion or Features

<p>This research investigated recognition memory for picture stories. Jenkins, Wald and Pittenger (1978) had found that when subjects viewed a slide sequence which depicted an every-day event, in a later recognition memory test they correctly rejected distractors which were inconsistent with the event but falsely accepted consistent distractors. Jenkins interpreted this result as evidence that fusion - the abstraction of visual events - determined memory performance. He argued that subjects compared the test slides to the abstracted event and accepted those which were consistent with the event. A series of experiments examined the possibility that performance was due not to fusion but to confusion with respect to the featural details of the stimulus material. This alternative interpretation argued that consistent slides had more features in common with acquisition slides than did the inconsistent slides and that the variables of semantic consistency and featural similarity had been confounded. The first experiment manipulated acquisition material and found that subjects who saw a disordered acquisition sequence falsely accepted consistent slides. The second experiment manipulated acquisition conditions and found that subjects who were inhibited from fusing the event by being required to perform a non-semantic task during acquisition falsely accepted consistent slides. Neither of these results supported a fusion interpretation since acceptance of consistent slides occurred under conditions where fusion of the event was not expected. The third experiment manipulated the test conditions and found that acceptance of both consistent and inconsistent slides was less likely with delayed tests although fusion of the event should have led to no change in the likelihood of accepting inconsistent slides. The fourth and fifth experiments re-examined the manipulation of presentation order and demonstrated that subjects were unable to reconstruct the event from a disordered sequence and yet still falsely accepted consistent slides. Each test of the fusion interpretation which had attempted to separate the variables of features and meaning indirectly had indicated that recognition performance was not due to abstraction of the visual event. A final experiment attempted to find explicit evidence for a featural interpretation of the results by directly varying featural similarity of consistent distractor slides to slides from the originally viewed sequence while keeping the degree of semantic consistency constant. Although this experiment failed to support a featural account, the converging evidence from all experiments indicated that recognition memory for picture stories is based to a large extent on the featural properties of the stimulus material. An account of performance solely in terms of visual abstraction is not adequate. Moreover, unless the variables of featural similarity and meaning can be separated directly in the test material, this recognition paradigm is unlikely to provide a means for examining the influence of schemata on recognition memory for picture stories.</p>

Recognition Memory for Pictorial Events: Fusion or Features

<p>This research investigated recognition memory for picture stories. Jenkins, Wald and Pittenger (1978) had found that when subjects viewed a slide sequence which depicted an every-day event, in a later recognition memory test they correctly rejected distractors which were inconsistent with the event but falsely accepted consistent distractors. Jenkins interpreted this result as evidence that fusion - the abstraction of visual events - determined memory performance. He argued that subjects compared the test slides to the abstracted event and accepted those which were consistent with the event. A series of experiments examined the possibility that performance was due not to fusion but to confusion with respect to the featural details of the stimulus material. This alternative interpretation argued that consistent slides had more features in common with acquisition slides than did the inconsistent slides and that the variables of semantic consistency and featural similarity had been confounded. The first experiment manipulated acquisition material and found that subjects who saw a disordered acquisition sequence falsely accepted consistent slides. The second experiment manipulated acquisition conditions and found that subjects who were inhibited from fusing the event by being required to perform a non-semantic task during acquisition falsely accepted consistent slides. Neither of these results supported a fusion interpretation since acceptance of consistent slides occurred under conditions where fusion of the event was not expected. The third experiment manipulated the test conditions and found that acceptance of both consistent and inconsistent slides was less likely with delayed tests although fusion of the event should have led to no change in the likelihood of accepting inconsistent slides. The fourth and fifth experiments re-examined the manipulation of presentation order and demonstrated that subjects were unable to reconstruct the event from a disordered sequence and yet still falsely accepted consistent slides. Each test of the fusion interpretation which had attempted to separate the variables of features and meaning indirectly had indicated that recognition performance was not due to abstraction of the visual event. A final experiment attempted to find explicit evidence for a featural interpretation of the results by directly varying featural similarity of consistent distractor slides to slides from the originally viewed sequence while keeping the degree of semantic consistency constant. Although this experiment failed to support a featural account, the converging evidence from all experiments indicated that recognition memory for picture stories is based to a large extent on the featural properties of the stimulus material. An account of performance solely in terms of visual abstraction is not adequate. Moreover, unless the variables of featural similarity and meaning can be separated directly in the test material, this recognition paradigm is unlikely to provide a means for examining the influence of schemata on recognition memory for picture stories.</p>

الفرائد الواردة في سياق الحديث عن الإعراض عن القرآن الكريم

يهدف البحث إلى: الكشف عن ظاهرة التمكّن الدلالي للفرائد القرآنية الواردة في سياق الحديث عن القرآن الكريم، وهي ظاهرة تعكس إعجازه البياني في أسلوبه، ولتحقيق هذا الهدف فقد سلك الباحث المنهج التحليلي اللغوي، والمنهج الوصفي، وخلصت الدراسة إلى أنّ ظاهرة التمكّن الدّلالي للفرائد القرآنية ترتبط بالعلاقات بين البُنى المتوالية للسورة -الواردة فيها الفريدة-بأكملها على وفق: مناسبة دلالة الفريدة معجميا، لسياقها الذي استدعاها، وهاتان العلاقتان مجتمعتان مكّنتا الفريدة دلاليا من التفرّد في سياقها بحيث تؤدّي المعنى المراد كله بدقة ولا يغني غيرُها من مرادفاتها غناءَها، ليثبت بذلك بناؤها واتّساقها الفني وفيضها المتجدد من الظلال والمعاني والدلالات التي تتّسع وتنمو وتفيض وتتجدّد بتدبّر هذه الفرائد، وأوصت الدراسة بمواصلة رَصْد ظاهرة التمكّن الدلالي للفرائد القرآنية الواردة في موضوعات أخرى، وتجديد قراءتنا للقرآن الكريم ومدارسته بما يحقق تدبّره. The purpose of the research is to highlight the phenomenon of the semantic mastery of the Quranic uniques in the context of the Holy Quran, a phenomenon that reflects its graphic miracles in its style and aims to reveal its significance to the purposes of the Holy Quran and its Surah's, to achieve the aim of the study: the researcher used the linguistic analytical method and the descriptive method, the study concluded in its results to phenomenon of the artistic semantic consistency of the Qur'anic authors is related to the relations between the successive structures of the surah - in which the unique - whole is presented according to: a unique and lexical connotation of the context in which they were called, the purpose of the sura and the purposes of the Qur'an, so that the meaning of the whole meaning accurately and does not enrich other synonyms singing, to prove the building and artistic consistency and its renewed flow of shadows and meanings and meanings that expand and grow and overflowing and renewed by meditating in these uniques, the study recommended the continuation of monitoring the phenomenon of semantic mastery of Quranic Uniques mentioned in other subjects and renewal of our reading of the Holy Quran and study it to achieve its meditation.

Characteristics of The Sentence in The Qur’anic Story

The holy Qur’an has adopted many methods in conveying its message to people, and among these methods is the style of the story that it has paid special attention to, form and content. The Quranic story has occupied a wide area in the Quranic context, and has actively contributed to its statement, rhetoric, miraculousness and verdicts, in addition to the entertainment it contains for the Messenger (may God bless him and grant him peace). From this standpoint, we will focus our research on the characteristics of the sentence in the context of the Qur’anic story, especially the stylistic characteristics. And what it achieves in terms of statement, significance and influence on the recipient's soul, in addition to achieving the mystery of the Quranic miracles, which is one of its goals for which it was revealed. We have examined these distinct characteristics of the sentence in the Qur’anic story through three axes or demands, namely: First: Verbal and semantic consistency. The second: the combination of compound and simple sentences. Third: the stylistic diversity between the nominative and the actual. This brief research discusses the most important advantages of the sentence in the context of the Quranic story. As a major unit in textual analysis, especially the sentences mentioned in the context of the Makkah story; Because of their importance and distinctive characteristics in the overall analysis of those stories, while not forgetting the civil story and indicating the semantic differences between them. The research reached many results, including that the Qur’an story in the Makkah surahs relied remarkably on the narration of events on short sentences. While the story in the Madinah surahs leaned towards long sentences, including that the Qur’an story relied on simple sentences, especially in dialogue stories, rather than on compound sentences, and also that the expression in actual sentences was overwhelming in the narrative use over nominal sentences what fits the actual sentence in its connotation of renewal, occurrence and movement, unlike the nominal one that denotes permanence and continuity.