Recent advances in molecular machine learning, especially deep neural networks such as Graph Neural Networks (GNNs) for predicting structure activity relationships (SAR) have shown tremendous potential in computer-aided drug discovery. However, the applicability of such deep neural networks are limited by the requirement of large amounts of training data. In order to cope with limited training data for a target task, transfer learning for SAR modeling has been recently adopted to leverage information from data of related tasks. In this work, in contrast to the popular parameter-based transfer learning such as pretraining, we develop novel deep transfer learning methods TAc and TAc-fc to leverage source domain data and transfer useful information to the target domain. TAc learns to generate effective molecular features that can generalize well from one domain to another, and increase the classification performance in the target domain. Additionally, TAc-fc extends TAc by incorporating novel components to selectively learn feature-wise and compound-wise transferability. We used the bioassay screening data from PubChem, and identified 120 pairs of bioassays such that the active compounds in each pair are more similar to each other compared to its inactive compounds. Overall, TAc achieves the best performance with average ROC-AUC of 0.801; it significantly improves ROC-AUC of 83% target tasks with average task-wise performance improvement of 7.102%, compared to the best baseline FCN-dmpna (DT). Our experiments clearly demonstrate that TAc achieves significant improvement over all baselines across a large number of target tasks. Furthermore, although TAc-fc achieves slightly worse ROC-AUC on average compared to TAc (0.798 vs 0.801), TAc-fc still achieves the best performance on more tasks in terms of PR-AUC and F1 compared to other methods.

Speech emotion conversion is the task of modifying the perceived emotion of a speech utterance while preserving the lexical content and speaker identity. In this study, we cast the problem of emotion conversion as a spoken language translation task. We decompose speech into discrete and disentangled learned representations, consisting of content units, F0, speaker, and emotion. First, we modify the speech content by translating the content units to a target emotion, and then predict the prosodic features based on these units. Finally, the speech waveform is generated by feeding the predicted representations into a neural vocoder. Such a paradigm allows us to go beyond spectral and parametric changes of the signal, and model non-verbal vocalizations, such as laughter insertion, yawning removal, etc. We demonstrate objectively and subjectively that the proposed method is superior to the baselines in terms of perceived emotion and audio quality. We rigorously evaluate all components of such a complex system and conclude with an extensive model analysis and ablation study to better emphasize the architectural choices, strengths and weaknesses of the proposed method. Samples and code will be publicly available under the following link: https://speechbot.github.io/emotion.

The unprecedented surge of data volume in wireless networks empowered with artificial intelligence (AI) opens up new horizons for providing ubiquitous data-driven intelligent services. Traditional cloud-centric machine learning (ML)-based services are implemented by collecting datasets and training models centrally. However, this conventional training technique encompasses two challenges: (i) high communication and energy cost due to increased data communication, (ii) threatened data privacy by allowing untrusted parties to utilise this information. Recently, in light of these limitations, a new emerging technique, coined as federated learning (FL), arose to bring ML to the edge of wireless networks. FL can extract the benefits of data silos by training a global model in a distributed manner, orchestrated by the FL server. FL exploits both decentralised datasets and computing resources of participating clients to develop a generalised ML model without compromising data privacy. In this article, we introduce a comprehensive survey of the fundamentals and enabling technologies of FL. Moreover, an extensive study is presented detailing various applications of FL in wireless networks and highlighting their challenges and limitations. The efficacy of FL is further explored with emerging prospective beyond fifth generation (B5G) and sixth generation (6G) communication systems. The purpose of this survey is to provide an overview of the state-of-the-art of FL applications in key wireless technologies that will serve as a foundation to establish a firm understanding of the topic. Lastly, we offer a road forward for future research directions.

Background and Objective: The Acetabular version, an essential factor in total hip arthroplasty, is measured by CT scan as the gold standard. The dose of radiation and expensiveness of CT make anterior-posterior pelvic radiograph an appropriate alternative procedure. In this study, we applied a deep learning approach on anteroposterior pelvic X-rays to measure anatomical version, eliminating the necessity of using Computed tomography scan. Methods: The right and left acetabular version angles of the hips of 300 patients are computed using their CT images. The proposed deep learning model, Attention on Pretrained-VGG16 for Bone Age, is applied to the AP images of the included population. The age and gender of these people are added as two other inputs to the last fully connected layer of attention mechanism. As the output, the angles of both hips are predicted. Results: The angles of hips computed on CT increase as people get older with the mean values of 16.54 and 16.11 (right and left angles) for men and 20.61 and 19.55 for women in our dataset. The predicted errors in the estimation of right and left angles using the proposed method of deep learning are in the accurate region of error (<=3 degrees) which shows the ability of the proposed method in measuring anatomical version based on AP images. Conclusion: The suggested algorithm, applying pre-trained vgg16 on the AP images of the pelvis of patients followed by an attention model considering age and gender of patients, can assess version accurately using only AP radiographs while obviating the need for CT scan. The applied technique of estimation of anatomical acetabular version based on AP pelvic images using DL approaches, to the best of authors' knowledge, has not been published yet.

Definitions are essential for term understanding. Recently, there is an increasing interest in extracting and generating definitions of terms automatically. However, existing approaches for this task are either extractive or abstractive - definitions are either extracted from a corpus or generated by a language generation model. In this paper, we propose to combine extraction and generation for definition modeling: first extract self- and correlative definitional information of target terms from the Web and then generate the final definitions by incorporating the extracted definitional information. Experiments demonstrate our framework can generate high-quality definitions for technical terms and outperform state-of-the-art models for definition modeling significantly.

An effective and efficient encoding of the source code of a computer program is critical to the success of sequence-to-sequence deep neural network models for tasks in computer program comprehension, such as automated code summarization and documentation. A significant challenge is to find a sequential representation that captures the structural/syntactic information in a computer program and facilitates the training of the learning models. In this paper, we propose to use the Pr\"ufer sequence of the Abstract Syntax Tree (AST) of a computer program to design a sequential representation scheme that preserves the structural information in an AST. Our representation makes it possible to develop deep-learning models in which signals carried by lexical tokens in the training examples can be exploited automatically and selectively based on their syntactic role and importance. Unlike other recently-proposed approaches, our representation is concise and lossless in terms of the structural information of the AST. Empirical studies on real-world benchmark datasets, using a sequence-to-sequence learning model we designed for code summarization, show that our Pr\"ufer-sequence-based representation is indeed highly effective and efficient, outperforming significantly all the recently-proposed deep-learning models we used as the baseline models.

Inherently, the legal domain contains a vast amount of data in text format. Therefore it requires the application of Natural Language Processing (NLP) to cater to the analytically demanding needs of the domain. The advancement of NLP is spreading through various domains, such as the legal domain, in forms of practical applications and academic research. Identifying critical sentences, facts and arguments in a legal case is a tedious task for legal professionals. In this research we explore the usage of sentence embeddings for multi-class classification to identify critical sentences in a legal case, in the perspective of the main parties present in the case. In addition, a task-specific loss function is defined in order to improve the accuracy restricted by the straightforward use of categorical cross entropy loss.

Digital Art created via Convolutional Neural Networks and Deep Learning. Style: Knife Painting Image: Seascape Original Image by ELG21@pixabay

We can't discuss Convolutional neural networks before skimming over convolution and pooling theory first. Both are simpler than you think, but extremely capable in image classification. Convolutional neural networks are a special type of neural network used for image classification. At the heart of any convolutional neural network lies convolution, an operation highly specialized at detecting patterns in images. Convolutional layers require you to specify the number of filters.

The most interesting development of the week had nothing to do with Facebook or even Google losing its appeal against a €2.4bn fine from the European commission for abusing its monopoly of search to the detriment of competitors to its shopping service. The bigger deal was that DeepMind, a London-based offshoot of Google (or, to be precise, its holding company, Alphabet) was moving into the pharmaceutical business via a new company called Isomorphic Labs, the goal of which is grandly described as "reimagining the entire drug discovery process from first principles with an AI-first approach". Since they're interested in first principles, let us first clarify that reference to AI. What it means in this context is not anything that is artificially intelligent, but simply machine learning, a technology of which DeepMind is an acknowledged master. AI has become a classic example of Orwellian newspeak adopted by the tech industry to sanitise a data-gobbling, energy-intensive technology that, like most things digital, has both socially useful and dystopian applications.