This paper compares 14 information retrieval metrics based on graded relevance, together with 10 traditional metrics based on binary relevance, in terms of stability, sensitivity and resemblance of ...system rankings. More specifically, we compare these metrics using the Buckley/Voorhees stability method, the Voorhees/Buckley swap method and Kendall’s rank correlation, with three data sets comprising test collections and submitted runs from NTCIR. Our experiments show that (Average) Normalised Discounted Cumulative Gain at document cut-off
l are the best among the
rank-based graded-relevance metrics, provided that
l is large. On the other hand, if one requires a
recall-based graded-relevance metric that is highly correlated with Average Precision, then Q-measure is the best choice. Moreover, these best graded-relevance metrics are at least as stable and sensitive as Average Precision, and are fairly robust to the choice of gain values.
Pre-trained language models are the cornerstone of modern natural language processing and information retrieval. However, fine-tuning all the parameters reduces the efficiency of models both in ...training and inference owing to their increasingly heavy structures. Existing methods for parameter efficiency still require approximately 1 MB of storage and have approximately 10 7 operations during model deployment and inference. This puts a strain on the storage and processor capacity of end devices such as smartphones and IoT equipment, and slow model inference adversely affecting the user experience. To achieve more efficient and storage-friendly inference compared to mainstream methods, such as low-rank adaptation (LoRA) and Adapter, LayerConnect (hyper-network-assisted interlayer connectors) is proposed in this paper. Extensive experiments were conducted to validate the performance of LayerConnect for two essential tasks with completely different learning frameworks and purposes: natural language understanding (using the general language understanding evaluation (GLUE) benchmark) and information retrieval (using the a contextualized inverted list (COIL) framework). For both tasks, our LayerConnect saves up to 95.31% and 91.18% of parameters in LoRA and Adapter, respectively. In contrast, LayerConnect maintains performance degradation for GLUE and COIL to less than 8% and 3%, compared to LoRA. When compared to Adapter, the numbers become 5% and 3%, for GLUE and COIL, respectively. In addition, LayerConnect required approximately 100 kB of storage per task-specific trained model for both tasks and reduced the number of operations in the model inference by four orders of magnitude, reaching approximately 10 3 .
Contrastive learning is a promising approach to unsupervised learning, as it inherits the advantages of well-studied deep models without a dedicated and complex model design. In this paper, based on ...bidirectional encoder representations from transformers (BERT) and long-short term memory (LSTM) neural networks, we propose self-supervised contrastive learning (SCL) as well as few-shot contrastive learning (FCL) with unsupervised data augmentation (UDA) for text clustering. BERT-SCL outperforms state-of-the-art unsupervised clustering approaches for short texts and for long texts in terms of several clustering evaluation measures. LSTM-SCL also shows good performance for short text clustering. BERT-FCL achieves performance close to supervised learning, and BERT-FCL with UDA further improves the performance for short texts. LSTM-FCL outperforms the supervised model in terms of several clustering evaluation measures. Our experiment results suggest that both SCL and FCL are effective for text clustering.
An integrated geophysical approach comprising microtremor chain array and self-potential surveys was used to assess the internal structure of landslide dams subject to possible piping erosion in ...selected sites in Japan and Kyrgyzstan. The non-invasive geophysical approach is cost effective, environmentally friendly and portable, and hence, it has proven to be valuable for the geotechnical assessment of landslide dams where piping can trigger failure of the dam. While the microtremor chain array survey results revealed the internal structure of the landslide dam, the self-potential survey results indicated the path of anomalous seepage zones. In the surveyed sites of long-existing landslide dams, the presence of a seepage path in the dam was confirmed by a good correlation between the areas of low phase velocity and large negative self-potential anomalies. In summary, this integrated geophysical approach could be useful for the early risk assessment of landslide dams and prediction of landslide dam failure by piping.
•A microtremor chain array survey was successfully used to detect the internal structure of landslide dams.•Self-potential surveys are useful for detecting the seepage route in a landslide dam.•The integration of the two surveys could be useful in evaluating landslide dams’ susceptibility to failure.
New record efficiencies have been achieved on Cd-free Cu(In,Ga)(Se,S)2 thin-film photovoltaic submodules prepared by a two-step sulfurization after selenization process. Aperture area efficiencies of ...19.2% and 19.8% were independently confirmed on a 30 cm × 30 cm submodule (841 cm 2 ) and on a 7 cm × 5 cm minimodule (24.2 cm 2 ), respectively. These achievements were brought about by transferring several key techniques, especially atomic layer deposited (Zn,Mg)O second buffer layer and K treatment of the absorber surface, from the fundamental study of small-area cell development. The former technique was applied to the submodule and both techniques were implemented into the minimodule. The (Zn,Mg)O second buffer layer increases transmittance of the window layer and improves junction quality resulting in the reduced interface recombination. The K treatment, which was developed by reference to the postdeposition treatment widely used in the co-evaporation process, significantly enhances open-circuit voltage and fill factor. Several material and device characterizations performed to illuminate the effects of the K treatment showed that increased free carrier concentration and reduced carrier recombination throughout the whole absorber film contributed to the improved performance. Contrary to the conventional postdeposition treatment in the co-evaporation process, significant depletion of Cu at the absorber surface was not observed, which can be attributed to S-rich circumstances of our absorber surface. The achievement of nearly 20% efficiency on the minimodule having identical structure to the production modules ensures further performance improvements in industrial Cu(In,Ga)(Se,S) 2 modules in the near future.
We evaluated plasma cell-free DNA (cfDNA) and tissue-based sequencing concordance for comprehensive oncogenic driver detection in non-small cell lung cancer (NSCLC) using a large-scale prospective ...screening cohort (LC-SCRUM-Liquid).
Blood samples were prospectively collected within 4 weeks of corresponding tumor tissue sampling from patients with advanced NSCLC to investigate plasma cfDNA sequencing concordance for alterations in 8 oncogenes (EGFR, KRAS, BRAF, HER2, MET, ALK, RET, and ROS1) compared with tissue-based next-generation targeted sequencing.
Paired blood and tissue samples were obtained in 1,062/1,112 enrolled patients with NSCLC. Oncogenic alteration was detected by plasma cfDNA sequencing and tissue assay in 455 (42.8%) and 537 (50.5%) patients, respectively. The positive percent agreement of plasma cfDNA sequencing compared with tissue DNA and RNA assays were 77% (EGFR, 78%; KRAS, 75%; BRAF, 85%; HER2, 72%) and 47% (ALK, 46%; RET, 57%; ROS1, 18%; MET, 66%), respectively. Oncogenic drivers were positive for plasma cfDNA and negative for tissue due to unsuccessful genomic analysis from poor-quality tissue samples (70%), and were negative for plasma cfDNA and positive for tissue due to low sensitivity of cfDNA analysis (61%). In patients with positive oncogenic drivers by plasma cfDNA sequencing but negative by tissue assay, the response rate of genotype-matched therapy was 85% and median progression-free survival was 12.7 months.
Plasma cfDNA sequencing in patients with advanced NSCLC showed relatively high sensitivity for detecting gene mutations but low sensitivity for gene fusions and MET exon 14 skipping. This may be an alternative only when tissue assay is unavailable due to insufficient DNA and RNA. See related commentary by Jacobsen Skanderup et al., p. 1381.
Lung cancer is one of the most aggressive tumour types. Targeted therapies stratified by oncogenic drivers have substantially improved therapeutic outcomes in patients with non-small-cell lung cancer ...(NSCLC)
. However, such oncogenic drivers are not found in 25-40% of cases of lung adenocarcinoma, the most common histological subtype of NSCLC
. Here we identify a novel fusion transcript of CLIP1 and LTK using whole-transcriptome sequencing in a multi-institutional genome screening platform (LC-SCRUM-Asia, UMIN000036871). The CLIP1-LTK fusion was present in 0.4% of NSCLCs and was mutually exclusive with other known oncogenic drivers. We show that kinase activity of the CLIP1-LTK fusion protein is constitutively activated and has transformation potential. Treatment of Ba/F3 cells expressing CLIP1-LTK with lorlatinib, an ALK inhibitor, inhibited CLIP1-LTK kinase activity, suppressed proliferation and induced apoptosis. One patient with NSCLC harbouring the CLIP1-LTK fusion showed a good clinical response to lorlatinib treatment. To our knowledge, this is the first description of LTK alterations with oncogenic activity in cancers. These results identify the CLIP1-LTK fusion as a target in NSCLC that could be treated with lorlatinib.
In the field of multimodal understanding and generation, tackling inherent uncertainties is essential for mitigating ambiguous interpretations across multiple targets. We introduce the Probability ...Distribution Encoder (PDE), a versatile, plug-and-play module that utilizes sequence-level and feature-level interactions to model these uncertainties as probabilistic distributions. Furthermore, we demonstrate its adaptability by seamlessly integrating PDE into established frameworks. Compared to previous methods, our probabilistic approach substantially enriches multimodal semantic understanding. In addition to specific tasks, the unlabeled data contains rich prior knowledge, especially multimodal uncertainties. However, current pre-training methods are designed based on point representations, which hinders the effective functioning of our distribution representations. Therefore, we incorporate this uncertainty modeling into three new pre-training strategies: Distribution-based Vision-Language Contrastive Learning (D-VLC), Distribution-based Masked Language Modeling (D-MLM), and Distribution-based Image-Text Matching (D-ITM). Empirical experiments show that our models achieve State-of-the-Art (SOTA) results in a range of downstream tasks, including image-text retrieval, visual question answering, visual reasoning, visual entailment and video captioning. Furthermore, the qualitative results reveal several superior properties conferred by our methods, such as improved semantic expressiveness over point representations, and the ability to generate diverse yet accurate predictions.
Neuropathic pain has a substantial effect on quality of life (QOL). The Japanese Society of Pain Clinicians (JSPC) has developed clinical guidelines of pharmacotherapy for neuropathic pain. These ...guidelines offer clarity on recommendations based on both the most recent scientific evidence and expert opinions. Understanding the concept, disease entity, and burden of neuropathic pain, as well as its screening and diagnosis are important steps before starting pharmacotherapy. As well as other guidelines, the guidelines propose several lines of pharmacotherapies in a step-wise manner. To name a few different points, our guidelines propose an extract from inflamed cutaneous tissue of rabbits inoculated with vaccinia virus, which has been found to be effective for post-herpetic neuralgia in Japan, as one of the second-line drugs. When prescribing opioid analgesics, proposed as the third-line drugs, for neuropathic pain, the guidelines recommend physicians continue evaluations on either abuse or addiction. The guidelines do not recommend concomitant use of nonsteroidal anti-inflammatory drugs and acetaminophen because of lack of clinical evidence of their efficacy. If patients do not respond well to pharmacotherapy, which is prescribed in a step-wise manner, other treatment strategies should be considered to improve patients’ activities of daily living and QOL.
Personalized recommendation has received a lot of attention as a highly practical research topic. However, existing recommender systems provide the recommendations with a generic statement such as ...“Customers who bought this item also bought…”. Explainable recommendation, which makes a user aware of why such items are recommended, is in demand. The goal of our research is to make the users feel as if they are receiving recommendations from their friends. To this end, we formulate a new challenging problem called personalized reason generation for explainable recommendation for songs in conversation applications and propose a solution that generates a natural language explanation of the reason for recommending a song to that particular user. For example, if the user is a student, our method can generate an output such as “Campus radio plays this song at noon every day, and I think it sounds wonderful,” which the student may find easy to relate to. In the offline experiments, through manual assessments, the gain of our method is statistically significant on the relevance to songs and personalization to users comparing with baselines. Large-scale online experiments show that our method outperforms manually selected reasons by 8.2% in terms of click-through rate. Evaluation results indicate that our generated reasons are relevant to songs and personalized to users, and they attract users to click the recommendations.