Yova Kementchedjhieva

Jiaang Li, Antonia Karamolegkou, Yova Kementchedjhieva et al.

Large language models (LLMs) have complicated internal dynamics, but induce representations of words and phrases whose geometry we can study. Human language processing is also opaque, but neural response measurements can provide (noisy) recordings of activation during listening or reading, from which we can extract similar representations of words and phrases. Here we study the extent to which the geometries induced by these representations, share similarities in the context of brain decoding. We find that the larger neural language models get, the more their representations are structurally similar to neural response measurements from brain imaging. Code is available at \url{https://github.com/coastalcph/brainlm}.

Shanu Kumar, Shubhanshu Khandelwal, Akhila Yesantarao Venkata et al.

Prompts tuned for accuracy often grow long, raising inference cost on every model call. The best accuracy-cost trade-off depends on the task and the budget, so prompt optimization is a search over the Pareto front of accuracy and prompt-token cost rather than for one prompt. The usual shortcut, collapsing the objectives into a weighted sum, fixes the trade-off weight before search and often recovers only a narrow region of the front, a failure we call scalarization collapse. We present CRAFT (Cost-aware Refinement And Front-aware Tuning), a Pareto-front prompt optimizer that treats target-LLM validation calls as the scarce resource and allocates them to candidates near the optimistic candidate front. Each round, complementary accuracy-oriented and cost-oriented generators propose edits, Pareto-gap acquisition spends the per-round validation budget, and NSGA-II retention keeps a spread-out population. Across six classification and reasoning benchmarks, CRAFT's retained fronts reach both high-accuracy and low-cost regions, while accuracy-only, cost-only, and weighted-sum baselines each concentrate in narrower regions. The accuracy-cost trade-off becomes a post-search choice, not a pre-search weight.

Rita Ramos, Bruno Martins, Desmond Elliott et al.

Recent advances in image captioning have focused on scaling the data and model size, substantially increasing the cost of pre-training and finetuning. As an alternative to large models, we present SmallCap, which generates a caption conditioned on an input image and related captions retrieved from a datastore. Our model is lightweight and fast to train, as the only learned parameters are in newly introduced cross-attention layers between a pre-trained CLIP encoder and GPT-2 decoder. SmallCap can transfer to new domains without additional finetuning and can exploit large-scale data in a training-free fashion since the contents of the datastore can be readily replaced. Our experiments show that SmallCap, trained only on COCO, has competitive performance on this benchmark, and also transfers to other domains without retraining, solely through retrieval from target-domain data. Further improvement is achieved through the training-free exploitation of diverse human-labeled and web data, which proves to be effective for a range of domains, including the nocaps benchmark, designed to test generalization to unseen visual concepts.

Jiaang Li, Yova Kementchedjhieva, Constanza Fierro et al.

Large-scale pretrained language models (LMs) are said to ``lack the ability to connect utterances to the world'' (Bender and Koller, 2020), because they do not have ``mental models of the world' '(Mitchell and Krakauer, 2023). If so, one would expect LM representations to be unrelated to representations induced by vision models. We present an empirical evaluation across four families of LMs (BERT, GPT-2, OPT and LLaMA-2) and three vision model architectures (ResNet, SegFormer, and MAE). Our experiments show that LMs partially converge towards representations isomorphic to those of vision models, subject to dispersion, polysemy and frequency. This has important implications for both multi-modal processing and the LM understanding debate (Mitchell and Krakauer, 2023).

Yong Cao, Yova Kementchedjhieva, Ruixiang Cui et al.

Building upon the considerable advances in Large Language Models (LLMs), we are now equipped to address more sophisticated tasks demanding a nuanced understanding of cross-cultural contexts. A key example is recipe adaptation, which goes beyond simple translation to include a grasp of ingredients, culinary techniques, and dietary preferences specific to a given culture. We introduce a new task involving the translation and cultural adaptation of recipes between Chinese and English-speaking cuisines. To support this investigation, we present CulturalRecipes, a unique dataset comprised of automatically paired recipes written in Mandarin Chinese and English. This dataset is further enriched with a human-written and curated test set. In this intricate task of cross-cultural recipe adaptation, we evaluate the performance of various methods, including GPT-4 and other LLMs, traditional machine translation, and information retrieval techniques. Our comprehensive analysis includes both automatic and human evaluation metrics. While GPT-4 exhibits impressive abilities in adapting Chinese recipes into English, it still lags behind human expertise when translating English recipes into Chinese. This underscores the multifaceted nature of cultural adaptations. We anticipate that these insights will significantly contribute to future research on culturally-aware language models and their practical application in culturally diverse contexts.

Haz Sameen Shahgir, Xiaofu Chen, Yu Fu et al.

Vision Language Models (VLMs) achieve impressive performance across a wide range of multimodal tasks. However, on some tasks that demand fine-grained visual perception, they often fail even when the required information is present in their internal representations. In this work, we demonstrate that this gap arises from their narrow training pipeline which focuses on moving visual information to the textual space. Consequently, VLMs can only reason about visual entities that can be mapped to known concepts in the language space, leaving vision-focused tasks such as visual correspondence and reasoning about novel visual entities poorly supported. As a result, VLMs are severely limited in several important multimodal capabilities because they rely on brittle, hallucinated textual descriptions of visual entities that they cannot map to textual representations. We verify this behavior through visual correspondence tasks, in which VLMs must detect matching entities between two images. Testing across semantic, shape, and face correspondence tasks, we find that VLMs perform much better when the relevant entities are nameable in language than when they are unnameable. Mechanistically, our Logit Lens analyses confirm that VLMs explicitly assign semantic labels to nameable entities and surface more unique corresponding tokens compared to unnameable entities. Furthermore, we show that teaching completely arbitrary names for unknown entities improves performance, yet task-specific finetuning yields even stronger generalization without relying on language priors. Our findings suggest that current VLM failures on visual tasks reflect learned shortcuts from their training, rather than a fundamental limitation of multimodal architectures.

Jiahui Geng, Yova Kementchedjhieva, Preslav Nakov et al.

Multimodal large language models (MLLMs) carry the potential to support humans in processing vast amounts of information. While MLLMs are already being used as a fact-checking tool, their abilities and limitations in this regard are understudied. Here is aim to bridge this gap. In particular, we propose a framework for systematically assessing the capacity of current multimodal models to facilitate real-world fact-checking. Our methodology is evidence-free, leveraging only these models' intrinsic knowledge and reasoning capabilities. By designing prompts that extract models' predictions, explanations, and confidence levels, we delve into research questions concerning model accuracy, robustness, and reasons for failure. We empirically find that (1) GPT-4V exhibits superior performance in identifying malicious and misleading multimodal claims, with the ability to explain the unreasonable aspects and underlying motives, and (2) existing open-source models exhibit strong biases and are highly sensitive to the prompt. Our study offers insights into combating false multimodal information and building secure, trustworthy multimodal models. To the best of our knowledge, we are the first to evaluate MLLMs for real-world fact-checking.

Karima Kadaoui, Hanin Atwany, Hamdan Al-Ali et al.

We introduce JEEM, a benchmark designed to evaluate Vision-Language Models (VLMs) on visual understanding across four Arabic-speaking countries: Jordan, The Emirates, Egypt, and Morocco. JEEM includes the tasks of image captioning and visual question answering, and features culturally rich and regionally diverse content. This dataset aims to assess the ability of VLMs to generalize across dialects and accurately interpret cultural elements in visual contexts. In an evaluation of five prominent open-source Arabic VLMs and GPT-4V, we find that the Arabic VLMs consistently underperform, struggling with both visual understanding and dialect-specific generation. While GPT-4V ranks best in this comparison, the model's linguistic competence varies across dialects, and its visual understanding capabilities lag behind. This underscores the need for more inclusive models and the value of culturally-diverse evaluation paradigms.

Xiaofu Chen, Israfel Salazar, Yova Kementchedjhieva

As interest grows in generating long, detailed image captions, standard evaluation metrics become increasingly unreliable. N-gram-based metrics though efficient, fail to capture semantic correctness. Representational Similarity (RS) metrics, designed to address this, initially saw limited use due to high computational costs, while today, despite advances in hardware, they remain unpopular due to low correlation to human judgments. Meanwhile, metrics based on large language models (LLMs) show strong correlation with human judgments, but remain too expensive for iterative use during model development. We introduce SPECS (Specificity-Enhanced CLIPScore), a reference-free RS metric tailored to long image captioning. SPECS modifies CLIP with a new objective that emphasizes specificity: rewarding correct details and penalizing incorrect ones. We show that SPECS matches the performance of open-source LLM-based metrics in correlation to human judgments, while being far more efficient. This makes it a practical alternative for iterative checkpoint evaluation during image captioning model development.Our code can be found at https://github.com/mbzuai-nlp/SPECS.

Constanza Fierro, Nicolas Garneau, Emanuele Bugliarello et al.

Facts are subject to contingencies and can be true or false in different circumstances. One such contingency is time, wherein some facts mutate over a given period, e.g., the president of a country or the winner of a championship. Trustworthy language models ideally identify mutable facts as such and process them accordingly. We create MuLan, a benchmark for evaluating the ability of English language models to anticipate time-contingency, covering both 1:1 and 1:N relations. We hypothesize that mutable facts are encoded differently than immutable ones, hence being easier to update. In a detailed evaluation of six popular large language models, we consistently find differences in the LLMs' confidence, representations, and update behavior, depending on the mutability of a fact. Our findings should inform future work on the injection of and induction of time-contingent knowledge to/from LLMs.

Jonas F. Lotz, Hendra Setiawan, Stephan Peitz et al.

Subword tokenization requires balancing computational efficiency and vocabulary coverage, which often leads to suboptimal performance on languages and scripts not prioritized during training. We propose to augment pretrained language models with a vocabulary-free encoder that generates input embeddings from text rendered as pixels. Through experiments on English-centric language models, we demonstrate that our approach substantially improves machine translation performance and facilitates effective cross-lingual transfer, outperforming tokenizer-based methods. Furthermore, we find that pixel-based representations outperform byte-level approaches and standard vocabulary expansion. Our approach enhances the multilingual capabilities of monolingual language models without extensive retraining and reduces decoding latency via input compression.

Raza Imam, Asif Hanif, Jian Zhang et al.

Recently, test-time adaptation has garnered attention as a method for tuning models without labeled data. The conventional modus operandi for adapting pre-trained vision-language models (VLMs) during test-time primarily focuses on tuning learnable prompts; however, this approach overlooks potential distribution shifts in the visual representations themselves. In this work, we address this limitation by introducing Test-Time Noise Tuning (TNT), a novel method for handling unpredictable shifts in the visual space. TNT leverages, for the first time, a noise adaptation strategy that optimizes learnable noise directly in the visual input space, enabling adaptive feature learning from a single test sample. We further introduce a novel approach for inter-view representation alignment by explicitly enforcing coherence in embedding distances, ensuring consistent feature representations across views. Combined with scaled logits and confident view selection at inference, TNT substantially enhances VLM generalization and calibration, achieving average gains of +7.38% on natural distributions benchmark and +0.80% on cross-dataset evaluations over zero-shot CLIP. These improvements lay a strong foundation for adaptive out-of-distribution handling.

George Ibrahim, Rita Ramos, Yova Kementchedjhieva

Multilingual vision-language models have made significant strides in image captioning, yet they still lag behind their English counterparts due to limited multilingual training data and costly large-scale model parameterization. Retrieval-augmented generation (RAG) offers a promising alternative by conditioning caption generation on retrieved examples in the target language, reducing the need for extensive multilingual training. However, multilingual RAG captioning models often depend on retrieved captions translated from English, which can introduce mismatches and linguistic biases relative to the source language. We introduce CONCAP, a multilingual image captioning model that integrates retrieved captions with image-specific concepts, enhancing the contextualization of the input image and grounding the captioning process across different languages. Experiments on the XM3600 dataset indicate that CONCAP enables strong performance on low- and mid-resource languages, with highly reduced data requirements. Our findings highlight the effectiveness of concept-aware retrieval augmentation in bridging multilingual performance gaps.

Patrick Amadeus Irawan, Erland Hilman Fuadi, Shanu Kumar et al.

Adapting pretrained language models (LMs) into vision-language models (VLMs) can degrade their native linguistic capability due to representation shift and cross-modal interference introduced during multimodal adaptation. Such loss is difficult to recover, even with targeted task-specific fine-tuning using standard objectives. Prior recovery approaches typically introduce additional modules that act as intermediate alignment layers to maintain or isolate modality-specific subspaces, which increases architectural complexity, adds parameters at inference time, and limits flexibility across models and settings. We propose LinguDistill, an adapter-free distillation method that restores linguistic capability by utilizing the original frozen LM as a teacher. We overcome the key challenge of enabling vision-conditioned teacher supervision by introducing layer-wise KV-cache sharing, which exposes the teacher to the student's multimodal representations without modifying the architecture of either model. We then selectively distill the teacher's strong linguistic signal on language-intensive data to recover language capability, while preserving the student's visual grounding on multimodal tasks. As a result, LinguDistill recovers $\sim$10% of the performance lost on language and knowledge benchmarks, while maintaining comparable performance on vision-heavy tasks. Our findings demonstrate that linguistic capability can be recovered without additional modules, providing an efficient and practical solution to modality-specific degradation in multimodal models.

Belal Shoer, Yova Kementchedjhieva

Scientific visual question answering poses significant challenges for vision-language models due to the complexity of scientific figures and their multimodal context. Traditional approaches treat the figure and accompanying text (e.g., questions and answer options) as separate inputs. EXAMS-V introduced a new paradigm by embedding both visual and textual content into a single image. However, even state-of-the-art proprietary models perform poorly on this setup in zero-shot settings, underscoring the need for task-specific fine-tuning. To address the scarcity of training data in this "text-in-image" format, we synthesize a new dataset by converting existing separate image-text pairs into unified images. Fine-tuning a small multilingual multimodal model on a mix of our synthetic data and EXAMS-V yields notable gains across 13 languages, demonstrating strong average improvements and cross-lingual transfer.

Israfel Salazar, Desmond Elliott, Yova Kementchedjhieva

Contrastive vision-language models (VLMs) have made significant progress in binding visual and textual information, but understanding long, dense captions remains an open challenge. We hypothesize that compositionality, the capacity to reason about object-attribute bindings and inter-object relationships, is key to understanding longer captions. In this paper, we investigate the interaction between compositionality and long-caption understanding, asking whether training for one property enhances the other. We train and evaluate a range of models that target each of these capabilities. Our results reveal a bidirectional relationship: compositional training improves performance on long-caption retrieval, and training on long captions promotes compositionality. However, these gains are sensitive to data quality and model design. We find that training on poorly structured captions, or with limited parameter updates, fails to support generalization. Likewise, strategies that aim at retaining general alignment, such as freezing positional embeddings, do not improve compositional understanding. Overall, we find that compositional understanding and long-caption understanding are intertwined capabilities that can be jointly learned through training on dense, grounded descriptions. Despite these challenges, we show that models trained on high-quality, long-caption data can achieve strong performance in both tasks, offering practical guidance for improving VLM generalization.

Abdelrahman Mohamed, Yova Kementchedjhieva

Despite significant advances in vision-language models (VLMs), image captioning often suffers from a lack of detail, with base models producing short, generic captions. This limitation persists even though VLMs are equipped with strong vision and language backbones. While supervised data and complex reward functions have been proposed to improve detailed image captioning, we identify a simpler underlying issue: a bias towards the end-of-sequence (EOS) token, which is introduced during cross-entropy training. We propose an unsupervised method to debias the model's tendency to predict the EOS token prematurely. By reducing this bias, we encourage the generation of longer, more detailed captions without the need for intricate reward functions or supervision. Our approach is straightforward, effective, and easily applicable to any pretrained model. We demonstrate its effectiveness through experiments with three VLMs and on three detailed captioning benchmarks. Our results show a substantial increase in caption length and relevant details, albeit with an expected increase in the rate of hallucinations.

Sho Takishita, Jay Gala, Abdelrahman Mohamed et al.

Many vision-language models (VLMs) that prove very effective at a range of multimodal task, build on CLIP-based vision encoders, which are known to have various limitations. We investigate the hypothesis that the strong language backbone in VLMs compensates for possibly weak visual features by contextualizing or enriching them. Using three CLIP-based VLMs, we perform controlled self-attention ablations on a carefully designed probing task. Our findings show that despite known limitations, CLIP visual representations offer ready-to-read semantic information to the language decoder. However, in scenarios of reduced contextualization in the visual representations, the language decoder can largely compensate for the deficiency and recover performance. This suggests a dynamic division of labor in VLMs and motivates future architectures that offload more visual processing to the language decoder.

Muhammad Huzaifa, Yova Kementchedjhieva

Text-to-image retrieval is a critical task for managing diverse visual content, but common benchmarks for the task rely on small, single-domain datasets that fail to capture real-world complexity. Pre-trained vision-language models tend to perform well with easy negatives but struggle with hard negatives--visually similar yet incorrect images--especially in open-domain scenarios. To address this, we introduce Episodic Few-Shot Adaptation (EFSA), a novel test-time framework that adapts pre-trained models dynamically to a query's domain by fine-tuning on top-k retrieved candidates and synthetic captions generated for them. EFSA improves performance across diverse domains while preserving generalization, as shown in evaluations on queries from eight highly distinct visual domains and an open-domain retrieval pool of over one million images. Our work highlights the potential of episodic few-shot adaptation to enhance robustness in the critical and understudied task of open-domain text-to-image retrieval.

David Romero, Chenyang Lyu, Haryo Akbarianto Wibowo et al.

Visual Question Answering (VQA) is an important task in multimodal AI, and it is often used to test the ability of vision-language models to understand and reason on knowledge present in both visual and textual data. However, most of the current VQA models use datasets that are primarily focused on English and a few major world languages, with images that are typically Western-centric. While recent efforts have tried to increase the number of languages covered on VQA datasets, they still lack diversity in low-resource languages. More importantly, although these datasets often extend their linguistic range via translation or some other approaches, they usually keep images the same, resulting in narrow cultural representation. To address these limitations, we construct CVQA, a new Culturally-diverse multilingual Visual Question Answering benchmark, designed to cover a rich set of languages and cultures, where we engage native speakers and cultural experts in the data collection process. As a result, CVQA includes culturally-driven images and questions from across 30 countries on four continents, covering 31 languages with 13 scripts, providing a total of 10k questions. We then benchmark several Multimodal Large Language Models (MLLMs) on CVQA, and show that the dataset is challenging for the current state-of-the-art models. This benchmark can serve as a probing evaluation suite for assessing the cultural capability and bias of multimodal models and hopefully encourage more research efforts toward increasing cultural awareness and linguistic diversity in this field.

Rustam Abdumalikov, Pasquale Minervini, Yova Kementchedjhieva

The performance of Open-Domain Question Answering (ODQA) retrieval systems can exhibit sub-optimal behavior, providing text excerpts with varying degrees of irrelevance. Unfortunately, many existing ODQA datasets lack examples specifically targeting the identification of irrelevant text excerpts. Previous attempts to address this gap have relied on a simplistic approach of pairing questions with random text excerpts. This paper aims to investigate the effectiveness of models trained using this randomized strategy, uncovering an important limitation in their ability to generalize to irrelevant text excerpts with high semantic overlap. As a result, we observed a substantial decrease in predictive accuracy, from 98% to 1%. To address this limitation, we discovered an efficient approach for training models to recognize such excerpts. By leveraging unanswerable pairs from the SQuAD 2.0 dataset, our models achieve a nearly perfect (~100%) accuracy when confronted with these challenging text excerpts.

Ilias Chalkidis, Yova Kementchedjhieva

Multi-label text classification (MLC) is a challenging task in settings of large label sets, where label support follows a Zipfian distribution. In this paper, we address this problem through retrieval augmentation, aiming to improve the sample efficiency of classification models. Our approach closely follows the standard MLC architecture of a Transformer-based encoder paired with a set of classification heads. In our case, however, the input document representation is augmented through cross-attention to similar documents retrieved from the training set and represented in a task-specific manner. We evaluate this approach on four datasets from the legal and biomedical domains, all of which feature highly skewed label distributions. Our experiments show that retrieval augmentation substantially improves model performance on the long tail of infrequent labels especially so for lower-resource training scenarios and more challenging long-document data scenarios.

Yova Kementchedjhieva, Ilias Chalkidis

Standard methods for multi-label text classification largely rely on encoder-only pre-trained language models, whereas encoder-decoder models have proven more effective in other classification tasks. In this study, we compare four methods for multi-label classification, two based on an encoder only, and two based on an encoder-decoder. We carry out experiments on four datasets -- two in the legal domain and two in the biomedical domain, each with two levels of label granularity -- and always depart from the same pre-trained model, T5. Our results show that encoder-decoder methods outperform encoder-only methods, with a growing advantage on more complex datasets and labeling schemes of finer granularity. Using encoder-decoder models in a non-autoregressive fashion, in particular, yields the best performance overall, so we further study this approach through ablations to better understand its strengths.

Yova Kementchedjhieva, Anders Søgaard

Online conversations can sometimes take a turn for the worse, either due to systematic cultural differences, accidental misunderstandings, or mere malice. Automatically forecasting derailment in public online conversations provides an opportunity to take early action to moderate it. Previous work in this space is limited, and we extend it in several ways. We apply a pretrained language encoder to the task, which outperforms earlier approaches. We further experiment with shifting the training paradigm for the task from a static to a dynamic one to increase the forecast horizon. This approach shows mixed results: in a high-quality data setting, a longer average forecast horizon can be achieved at the cost of a small drop in F1; in a low-quality data setting, however, dynamic training propagates the noise and is highly detrimental to performance.

Yova Kementchedjhieva, Mark Anderson, Anders Søgaard

Some interpersonal verbs can implicitly attribute causality to either their subject or their object and are therefore said to carry an implicit causality (IC) bias. Through this bias, causal links can be inferred from a narrative, aiding language comprehension. We investigate whether pre-trained language models (PLMs) encode IC bias and use it at inference time. We find that to be the case, albeit to different degrees, for three distinct PLM architectures. However, causes do not always need to be implicit -- when a cause is explicitly stated in a subordinate clause, an incongruent IC bias associated with the verb in the main clause leads to a delay in human processing. We hypothesize that the temporary challenge humans face in integrating the two contradicting signals, one from the lexical semantics of the verb, one from the sentence-level semantics, would be reflected in higher error rates for models on tasks dependent on causal links. The results of our study lend support to this hypothesis, suggesting that PLMs tend to prioritize lexical patterns over higher-order signals.

Yova Kementchedjhieva, Di Lu, Joel Tetreault

News articles, image captions, product reviews and many other texts mention people and organizations whose name recognition could vary for different audiences. In such cases, background information about the named entities could be provided in the form of an appositive noun phrase, either written by a human or generated automatically. We expand on the previous work in appositive generation with a new, more realistic, end-to-end definition of the task, instantiated by a dataset that spans four languages (English, Spanish, German and Polish), two entity types (person and organization) and two domains (Wikipedia and News). We carry out an extensive analysis of the data and the task, pointing to the various modeling challenges it poses. The results we obtain with standard language generation methods show that the task is indeed non-trivial, and leaves plenty of room for improvement.

Gözde Gül Şahin, Yova Kementchedjhieva, Phillip Rust et al.

Deep neural models have repeatedly proved excellent at memorizing surface patterns from large datasets for various ML and NLP benchmarks. They struggle to achieve human-like thinking, however, because they lack the skill of iterative reasoning upon knowledge. To expose this problem in a new light, we introduce a challenge on learning from small data, PuzzLing Machines, which consists of Rosetta Stone puzzles from Linguistic Olympiads for high school students. These puzzles are carefully designed to contain only the minimal amount of parallel text necessary to deduce the form of unseen expressions. Solving them does not require external information (e.g., knowledge bases, visual signals) or linguistic expertise, but meta-linguistic awareness and deductive skills. Our challenge contains around 100 puzzles covering a wide range of linguistic phenomena from 81 languages. We show that both simple statistical algorithms and state-of-the-art deep neural models perform inadequately on this challenge, as expected. We hope that this benchmark, available at https://ukplab.github.io/PuzzLing-Machines/, inspires further efforts towards a new paradigm in NLP---one that is grounded in human-like reasoning and understanding.

Hila Gonen, Yova Kementchedjhieva, Yoav Goldberg

Many natural languages assign grammatical gender also to inanimate nouns in the language. In such languages, words that relate to the gender-marked nouns are inflected to agree with the noun's gender. We show that this affects the word representations of inanimate nouns, resulting in nouns with the same gender being closer to each other than nouns with different gender. While "embedding debiasing" methods fail to remove the effect, we demonstrate that a careful application of methods that neutralize grammatical gender signals from the words' context when training word embeddings is effective in removing it. Fixing the grammatical gender bias yields a positive effect on the quality of the resulting word embeddings, both in monolingual and cross-lingual settings. We note that successfully removing gender signals, while achievable, is not trivial to do and that a language-specific morphological analyzer, together with careful usage of it, are essential for achieving good results.

Yova Kementchedjhieva, Mareike Hartmann, Anders Søgaard

The task of bilingual dictionary induction (BDI) is commonly used for intrinsic evaluation of cross-lingual word embeddings. The largest dataset for BDI was generated automatically, so its quality is dubious. We study the composition and quality of the test sets for five diverse languages from this dataset, with concerning findings: (1) a quarter of the data consists of proper nouns, which can be hardly indicative of BDI performance, and (2) there are pervasive gaps in the gold-standard targets. These issues appear to affect the ranking between cross-lingual embedding systems on individual languages, and the overall degree to which the systems differ in performance. With proper nouns removed from the data, the margin between the top two systems included in the study grows from 3.4% to 17.2%. Manual verification of the predictions, on the other hand, reveals that gaps in the gold standard targets artificially inflate the margin between the two systems on English to Bulgarian BDI from 0.1% to 6.7%. We thus suggest that future research either avoids drawing conclusions from quantitative results on this BDI dataset, or accompanies such evaluation with rigorous error analysis.

Clara Vania, Yova Kementchedjhieva, Anders Søgaard et al.

Parsers are available for only a handful of the world's languages, since they require lots of training data. How far can we get with just a small amount of training data? We systematically compare a set of simple strategies for improving low-resource parsers: data augmentation, which has not been tested before; cross-lingual training; and transliteration. Experimenting on three typologically diverse low-resource languages---North Sámi, Galician, and Kazah---We find that (1) when only the low-resource treebank is available, data augmentation is very helpful; (2) when a related high-resource treebank is available, cross-lingual training is helpful and complements data augmentation; and (3) when the high-resource treebank uses a different writing system, transliteration into a shared orthographic spaces is also very helpful.

Johannes Bjerva, Yova Kementchedjhieva, Ryan Cotterell et al.

The study of linguistic typology is rooted in the implications we find between linguistic features, such as the fact that languages with object-verb word ordering tend to have post-positions. Uncovering such implications typically amounts to time-consuming manual processing by trained and experienced linguists, which potentially leaves key linguistic universals unexplored. In this paper, we present a computational model which successfully identifies known universals, including Greenberg universals, but also uncovers new ones, worthy of further linguistic investigation. Our approach outperforms baselines previously used for this problem, as well as a strong baseline from knowledge base population.

Johannes Bjerva, Yova Kementchedjhieva, Ryan Cotterell et al.

In the principles-and-parameters framework, the structural features of languages depend on parameters that may be toggled on or off, with a single parameter often dictating the status of multiple features. The implied covariance between features inspires our probabilisation of this line of linguistic inquiry---we develop a generative model of language based on exponential-family matrix factorisation. By modelling all languages and features within the same architecture, we show how structural similarities between languages can be exploited to predict typological features with near-perfect accuracy, outperforming several baselines on the task of predicting held-out features. Furthermore, we show that language embeddings pre-trained on monolingual text allow for generalisation to unobserved languages. This finding has clear practical and also theoretical implications: the results confirm what linguists have hypothesised, i.e.~that there are significant correlations between typological features and languages.

Yova Kementchedjhieva, Johannes Bjerva, Isabelle Augenstein

This paper documents the Team Copenhagen system which placed first in the CoNLL--SIGMORPHON 2018 shared task on universal morphological reinflection, Task 2 with an overall accuracy of 49.87. Task 2 focuses on morphological inflection in context: generating an inflected word form, given the lemma of the word and the context it occurs in. Previous SIGMORPHON shared tasks have focused on context-agnostic inflection---the "inflection in context" task was introduced this year. We approach this with an encoder-decoder architecture over character sequences with three core innovations, all contributing to an improvement in performance: (1) a wide context window; (2) a multi-task learning approach with the auxiliary task of MSD prediction; (3) training models in a multilingual fashion.

Mareike Hartmann, Yova Kementchedjhieva, Anders Søgaard

This paper presents a challenge to the community: Generative adversarial networks (GANs) can perfectly align independent English word embeddings induced using the same algorithm, based on distributional information alone; but fails to do so, for two different embeddings algorithms. Why is that? We believe understanding why, is key to understand both modern word embedding algorithms and the limitations and instability dynamics of GANs. This paper shows that (a) in all these cases, where alignment fails, there exists a linear transform between the two embeddings (so algorithm biases do not lead to non-linear differences), and (b) similar effects can not easily be obtained by varying hyper-parameters. One plausible suggestion based on our initial experiments is that the differences in the inductive biases of the embedding algorithms lead to an optimization landscape that is riddled with local optima, leading to a very small basin of convergence, but we present this more as a challenge paper than a technical contribution.

Yova Kementchedjhieva, Adam Lopez

Character language models have access to surface morphological patterns, but it is not clear whether or how they learn abstract morphological regularities. We instrument a character language model with several probes, finding that it can develop a specific unit to identify word boundaries and, by extension, morpheme boundaries, which allows it to capture linguistic properties and regularities of these units. Our language model proves surprisingly good at identifying the selectional restrictions of English derivational morphemes, a task that requires both morphological and syntactic awareness. Thus we conclude that, when morphemes overlap extensively with the words of a language, a character language model can perform morphological abstraction.

Yova Kementchedjhieva, Sebastian Ruder, Ryan Cotterell et al.

Most recent approaches to bilingual dictionary induction find a linear alignment between the word vector spaces of two languages. We show that projecting the two languages onto a third, latent space, rather than directly onto each other, while equivalent in terms of expressivity, makes it easier to learn approximate alignments. Our modified approach also allows for supporting languages to be included in the alignment process, to obtain an even better performance in low resource settings.

Sebastian Ruder, Ryan Cotterell, Yova Kementchedjhieva et al.

We introduce a novel discriminative latent variable model for bilingual lexicon induction. Our model combines the bipartite matching dictionary prior of Haghighi et al. (2008) with a representation-based approach (Artetxe et al., 2017). To train the model, we derive an efficient Viterbi EM algorithm. We provide empirical results on six language pairs under two metrics and show that the prior improves the induced bilingual lexicons. We also demonstrate how previous work may be viewed as a similarly fashioned latent-variable model, albeit with a different prior.