Tabular data is one of the most ubiquitous sources of information worldwide, spanning a wide variety of domains. This inherent heterogeneity has slowed the development of Tabular Foundation Models (TFMs) capable of fast generalization to unseen datasets. In-Context Learning (ICL) has recently emerged as a promising solution for TFMs, enabling dynamic adaptation to new tasks without additional tuning. While many studies have attempted to re-purpose large language models for tabular ICL, they have had limited success, so recent works have focused on developing tabular-specific foundation models. In this work, we propose an approach to combine ICL-based retrieval with self supervised learning to train tabular foundation models. We also investigate the utility of real vs. synthetic data for model pre-training, and show that real data can contain useful signal not easily captured in synthetic training. Specifically, we show that incorporating real data during the pre-training phase can lead to significantly faster training and better downstream generalization to unseen data. Our resulting model, **TabDPT**, achieves strong performance on both regression (CTR23) and classification (CC18) benchmarks. Importantly, we also demonstrate that with our pre-training procedure, scaling both model and data size leads to consistent performance improvements that follow power laws. This echoes scaling laws in LLMs and other foundation models, and suggests that large-scale TFMs can be achievable. We open-source our full pipeline: inference code including trained model weights can be found [here](https://github.com/layer6ai-labs/TabDPT-inference), and the training code to reproduce experiments can be found [here](https://github.com/layer6ai-labs/TabDPT-training).

The goal of multimodal alignment is to learn a single latent space that is shared between multimodal inputs. The most powerful models in this space have been trained using massive datasets of paired inputs and large-scale computational resources, making them prohibitively expensive to train in many practical scenarios. We surmise that existing unimodal encoders pre-trained on large amounts of unimodal data should provide an effective bootstrap to create multimodal models from unimodal ones at much lower costs. We therefore propose FuseMix, a multimodal augmentation scheme that operates on the latent spaces of arbitrary pre-trained unimodal encoders. Using FuseMix for multimodal alignment, we achieve competitive performance - and in certain cases outperform state-of-the art methods - in both image-text and audio-text retrieval, with orders of magnitude less compute and data: for example, we outperform CLIP on the Flickr30K text-to-image retrieval task with ~600 × fewer GPU days and ~80 × fewer image-text pairs. Additionally, we show how our method can be applied to convert pre-trained text-to-image generative models into audio-to-image ones. Code is available at: https://github.com/layer6ai-labs/fusemix.

Self-supervised representation learning (SSRL) has advanced considerably by exploiting the transformation invariance assumption under artificially designed data augmentations. While augmentation-based SSRL algorithms push the boundaries of performance in computer vision and natural language processing, they are often not directly applicable to other data modalities, and can conflict with application-specific data augmentation constraints. This paper presents an SSRL approach that can be applied to any data modality and network architecture because it does not rely on augmentations or masking. Specifically, we show that high-quality data representations can be learned by reconstructing random data projections. We evaluate the proposed approach on a wide range of representation learning tasks that span diverse modalities and real-world applications. We show that it outperforms multiple state-of-the-art SSRL baselines. Due to its wide applicability and strong empirical results, we argue that learning from randomness is a fruitful research direction worthy of attention and further study.

Tabular data is a pervasive modality spanning a wide range of domains, and this inherent diversity poses a considerable challenge for deep learning. Recent advancements using transformer-based in-context learning have shown promise on smaller and less complex tabular datasets, but have struggled to scale to larger and more complex ones. To address this limitation, we propose a combination of retrieval and fine-tuning: we can adapt the transformer to a local subset of the data by collecting nearest neighbours, and then perform task-specific fine-tuning with this retrieved set of neighbours in context. Using TabPFN as the base model -- currently the best tabular in-context learner -- and applying our retrieval and fine-tuning scheme on top results in what we call a locally-calibrated PFN, or LoCalPFN. We conduct extensive evaluation on 95 datasets curated by TabZilla from OpenML, upon which we establish a new state-of-the-art with LoCalPFN -- even with respect to tuned tree-based models. Notably, we show a significant boost in performance compared to the base in-context model, demonstrating the efficacy of our approach and advancing the frontier of deep learning in tabular data.

In text-video retrieval, the objective is to learn a cross-modal similarity function between a text and a video that ranks relevant text-video pairs higher than irrelevant pairs. However, videos inherently express a much wider gamut of information than texts. Instead, texts often capture sub-regions of entire videos and are most semantically similar to certain frames within videos. Therefore, for a given text, a retrieval model should focus on the text's most semantically similar video sub-regions to make a more relevant comparison. Yet, most existing works aggregate entire videos with-out directly considering text. Common text-agnostic ag-gregations schemes include mean-pooling or self-attention over the frames, but these are likely to encode misleading vi-sual information not described in the given text. To address this, we propose a cross-modal attention model called X-Pool that reasons between a text and the frames of a video. Our core mechanism is a scaled dot product attention for a text to attend to its most semantically similar frames. We then generate an aggregated video representation conditioned on the text's attention weights over the frames. We evaluate our method on three benchmark datasets of MSR-VTT, MSVD and LSMDC, achieving new state-of-the-art re-sults by up to 12% in relative improvement in Recall@ 1. Our findings thereby highlight the importance of joint text-video reasoning to extract important visual cues according to text. Full code and demo can be found at: layer6ai-labs.github.iolxpooll.

The majority of recent work in latent Collaborative Filtering (CF) has focused on developing new model architectures to learn accurate user and item representations. Typically, a standard pairwise loss function (BPR, Triplet, etc.) is used in these models, and little exploration is done on how to optimally extract signals from the available preference information. In the implicit setting, negative examples are sampled, and these losses allocate weights that solely depend on the difference in user distance between observed (positive) and negative item pairs. This can ignore valuable global information from other users and items, and lead to sub-optimal results. Motivated by this problem, we propose a novel loss which first leverages mining to select the most informative pairs, followed by a weighing process to allocate more weight to harder examples. Our weighting process consists of four different components, and incorporates distance information from other users, enabling the model to better position the learned representations. We conduct extensive experiments and demonstrate that our loss can be applied to different types of CF models leading to significant gains with each type. In particular, by applying our loss to the graph convolutional architecture, we achieve new state-of-the-art results on four different datasets. Further analysis shows that through our loss the model is able to learn better user-item representation space compared to other losses. Full code for this work is available here: https://github.com/layer6ai-labs/MCL.

Localizing actions in video is a core task in computer vision. The weakly supervised temporal localization problem investigates whether this task can be adequately solved with only video-level labels, significantly reducing the amount of expensive and error-prone annotation that is required. A common approach is to train a frame-level classifier where frames with the highest class probability are selected to make a video-level prediction. Frame-level activations are then used for localization. However, the absence of frame-level annotations cause the classifier to impart class bias on every frame. To address this, we propose the Action Selection Learning (ASL) approach to capture the general concept of action, a property we refer to as "actionness". Under ASL, the model is trained with a novel class-agnostic task to predict which frames will be selected by the classifier. Empirically, we show that ASL outperforms leading baselines on two popular bench-marks THUMOS-14 and ActivityNet-1.2, with 10.3% and 5.7% relative improvement respectively. We further analyze the properties of ASL and demonstrate the importance of actionness. Full code for this work is available here: https://github.com/layer6ai-labs/ASL.

Scene graph generation is an important task in computer vision aimed at improving the semantic understanding of the visual world. In this task, the model needs to detect objects and predict visual relationships between them. Most of the existing models predict relationships in parallel assuming their independence. While there are different ways to capture these dependencies, we explore a conditional approach motivated by the sequence-to-sequence (Seq2Seq) formalism. Different from the previous research, our proposed model predicts visual relationships one at a time in an autoregressive manner by explicitly conditioning on the already predicted relationships. Drawing from translation models in NLP, we propose an encoder-decoder model built using Transformers where the encoder captures global context and long range interactions. The decoder then makes sequential predictions by conditioning on the scene graph constructed so far. In addition, we introduce a novel reinforcement learning-based training strategy tailored to Seq2Seq scene graph generation. By using a self-critical policy gradient training approach with Monte Carlo search we directly optimize for the (mean) recall metrics and bridge the gap between training and evaluation. Experimental results on two public benchmark datasets demonstrate that our Seq2Seq learning approach achieves strong empirical performance, outperforming previous state-of-the-art, while remaining efficient in terms of training and inference time. Full code for this work is available here: https://github.com/layer6ai-labs/SGG-Seq2Seq.

Hyperbolic spaces offer a rich setup to learn embeddings with superior properties that have been leveraged in areas such as computer vision, natural language processing and computational biology. Recently, several hyperbolic approaches have been proposed to learn robust representations for users and items in the recommendation setting. However, these approaches don’t capture the higher order relationships that typically exist in the recommendation domain. Graph convolutional neural networks (GCNs) on the other hand excel at capturing higher order information by applying multiple levels of aggregation to local representations. In this paper we combine these frameworks in a novel way, by proposing a hyperbolic GCN model for collaborative filtering. We demonstrate that our model can be effectively learned with a margin ranking loss, and show that hyperbolic space has desirable properties under the rank margin setting. At test time, inference in our model is done using the hyperbolic distance which preserves the structure of the learned space. We conduct extensive empirical analysis on three public benchmarks and compare against a large set of baselines. Our approach achieves highly competitive results and outperforms leading baselines including the Euclidean GCN counterpart. We further study the properties of the learned hyperbolic embeddings and show that they offer meaningful insights into the data. Full code for this work is available here: https://github.com/layer6ai-labs/HGCF.

The Transformer architecture has achieved considerable success recently; the key component of the Transformer is the attention layer that enables the model to focus on important regions within an input sequence. Gradient optimization with attention layers can be notoriously difficult requiring tricks such as learning rate warmup to prevent divergence. As Transformer models are becoming larger and more expensive to train, recent research has focused on understanding and improving optimization in these architectures. In this work our contributions are two-fold: we first investigate and empirically validate the source of optimization problems in the encoder-decoder Transformer architecture; we then propose a new weight initialization scheme with theoretical justification, that enables training without warmup or layer normalization. Empirical results on public machine translation benchmarks show that our approach achieves leading accuracy, allowing to train deep Transformer models with 200 layers in both encoder and decoder (over 1000 attention/MLP blocks) without difficulty. Code for this work is available here: \url{https://github.com/layer6ai-labs/T-Fixup}.

Collaborative filtering with implicit feedback is a ubiquitous class of recommendation problems where only positive interactions such as purchases or clicks are observed. Autoencoder-based recommendation models have shown strong performance on many implicit feedback benchmarks. However, these models tend to suffer from popularity bias making recommendations less personalized. User-generated reviews contain a rich source of preference information, often with specific details that are important to each user, and can help mitigate the popularity bias. Since not all reviews are equally useful, existing work has been exploring various forms of attention to distill relevant information. In the majority of proposed approaches, representations from implicit feedback and review branches are simply concatenated at the end to generate predictions. This can prevent the model from learning deeper correlations between the two modalities and affect prediction accuracy. To address these problems, we propose a novel Two-headed Attention Fused Autoencoder (TAFA) model that jointly learns representations from user reviews and implicit feedback to make recommendations. We apply early and late modality fusion which allows the model to fully correlate and extract relevant information from both input sources. To further combat popularity bias, we leverage the Noise Contrastive Estimation (NCE) objective to “de-popularize” the fused user representation via a two-headed decoder architecture. Empirically, we show that TAFA outperforms leading baselines on multiple real-world benchmarks. Moreover, by tracing attention weights back to reviews we can provide explanations for the generated recommendations and gain further insights into user preferences. Full code for this work is available here: https://github.com/layer6ai-labs/TAFA.

We propose a novel graph-based approach for image retrieval. Given a nearest neighbor graph produced by the global descriptor model, we traverse it by alternating between exploit and explore steps. The exploit step maximally utilizes the immediate neighborhood of each vertex, while the explore step traverses vertices that are farther away in the descriptor space. By combining these two steps we can better capture the underlying image manifold, and successfully retrieve relevant images that are visually dissimilar to the query. Our traversal algorithm is conceptually simple, has few tunable parameters and can be implemented with basic data structures. This enables fast real-time inference for previously unseen queries with minimal memory overhead. Despite relative simplicity, we show highly competitive results on multiple public benchmarks, including the largest image retrieval dataset that is currently publicly available. Full code for this work is available here: https://github.com/layer6ai-labs/egt.

Despite recent progress in computer vision, image retrieval remains a challenging open problem. Numerous variations such as view angle, lighting and occlusion make it difficult to design models that are both robust and efficient. Many leading methods traverse the nearest neighbor graph to exploit higher order neighbor information and uncover the highly complex underlying manifold. In this work we propose a different approach where we leverage graph convolutional networks to directly encode neighbor information into image descriptors. We further leverage ideas from clustering and manifold learning, and introduce an unsupervised loss based on pairwise separation of image similarities. Empirically, we demonstrate that our model is able to successfully learn a new descriptor space that significantly improves retrieval accuracy, while still allowing efficient inner product inference. Experiments on five public benchmarks show highly competitive performance with up to 24\% relative improvement in mAP over leading baselines. Full code for this work is available here: https://github.com/layer6ai-labs/GSS.

Previous highly scalable One-Class Collaborative Filtering (OC-CF) methods such as Projected Linear Recommendation (PLRec) have advocated using fast randomized SVD to embed items into a latent space, followed by linear regression methods to learn personalized recommendation models per user. However, naive SVD embedding methods often exhibit a strong popularity bias that prevents them from accurately embedding less popular items, which is exacerbated by the extreme sparsity of implicit feedback matrices in the OC-CF setting. To address this deficiency, we leverage insights from Noise Contrastive Estimation (NCE) to derive a closed-form, efficiently computable "depopularized" embedding. We show that NCE item embeddings combined with a personalized user model from PLRec produces superior recommendations that adequately account for popularity bias. Further analysis of the popularity distribution of recommended items demonstrates that NCE-PLRec uniformly distributes recommendations over the popularity spectrum while other methods exhibit distinct biases towards specific popularity subranges. Empirically, NCE-PLRec produces highly competitive performance with run-times an order of magnitude faster than existing state-of-the-art approaches for OC-CF.

Latent models have become the default choice for recommender systems due to their performance and scalability. However, research in this area has primarily focused on modeling user-item interactions, and few latent models have been developed for cold start. Deep learning has recently achieved remarkable success showing excellent results for diverse input types. Inspired by these results we propose a neural network based latent model called DropoutNet to address the cold start problem in recommender systems. Unlike existing approaches that incorporate additional content-based objective terms, we instead focus on the optimization and show that neural network models can be explicitly trained for cold start through dropout. Our model can be applied on top of any existing latent model effectively providing cold start capabilities, and full power of deep architectures. Empirically we demonstrate state-of-the-art accuracy on publicly available benchmarks. Code is available at https://github.com/layer6ai-labs/DropoutNet.

In many collaborative filtering (CF) applications, latent approaches are the preferred model choice due to their ability to generate real-time recommendations efficiently. However, the majority of existing latent models are not designed for implicit binary feedback (views, clicks, plays etc.) and perform poorly on data of this type. Developing accurate models from implicit feedback is becoming increasingly important in CF since implicit feedback can often be collected at lower cost and in much larger quantities than explicit preferences. The need for accurate latent models for implicit data was further emphasized by the recently conducted Million Song Dataset Challenge organized by Kaggle. In this challenge, the results for the best latent model were orders of magnitude worse than neighbor-based approaches, and all the top performing teams exclusively used neighbor-based models. We address this problem and propose a new latent approach for binary feedback in CF. In our model, neighborhood similarity information is used to guide latent factorization and derive accurate latent representations. We show that even with simple factorization methods like SVD, our approach outperforms existing models and produces state-of-the-art results.

In declarative data cleaning, data semantics are encoded as constraints and errors arise when the data violates the constraints. Various forms of statistical and logical inference can be used to reason about and repair inconsistencies (errors) in data. Recently, unified approaches that repair both errors in data and errors in semantics (the constraints) have been proposed. However, both data-only approaches and unified approaches are by and large static in that they apply cleaning to a single snapshot of the data and constraints. We introduce a continuous data cleaning framework that can be applied to dynamic data and constraint environments. Our approach permits both the data and its semantics to evolve and suggests repairs based on the accumulated evidence to date. Importantly, our approach uses not only the data and constraints as evidence, but also considers the past repairs chosen and applied by a user (user repair preferences). We introduce a repair classifier that predicts the type of repair needed to resolve an inconsistency, and that learns from past user repair preferences to recommend more accurate repairs in the future. Our evaluation shows that our techniques achieve high prediction accuracy and generate high quality repairs. Of independent interest, our work makes use of a set of data statistics that are shown to be sensitive to predicting particular repair types.

We develop a flexible Conditional Random Field framework for supervised preference aggregation, which combines preferences from multiple experts over items to form a distribution over rankings. The distribution is based on an energy comprised of unary and pairwise potentials allowing us to effectively capture correlations between both items and experts. We describe procedures for learning in this modelnand demonstrate that inference can be done much more efficiently thannin analogous models. Experiments on benchmark tasks demonstrate significant performance gains over existing rank aggregation methods.

We present a general treatment of the problem of aggregating preferences from several experts into a consensus ranking, in the context where information about a target ranking is available. Specifically, we describe how such problems can be converted into a standard learning-to-rank one on which existing learning solutions can be invoked. This transformation allows us to optimize the aggregating function for any target IR metric, such as Normalized Discounted Cumulative Gain, or Expected Reciprocal Rank. When applied to crowdsourcing and meta-search benchmarks, our new algorithm improves on state-of-the-art preference aggregation methods.

Bipartite matching problems characterize many situations, ranging from ranking in information retrieval to correspondence in vision. Exact inference in real-world applications of these problems is intractable, making efficient approximation methods essential for learning and inference. In this paper we propose a novel {\it sequential matching} sampler based on the generalization of the Plackett-Luce model, which can effectively make large moves in the space of matchings. This allows the sampler to match the difficult target distributions common in these problems: highly multimodal distributions with well separated modes. We present experimental results with bipartite matching problems - ranking and image correspondence - which show that the sequential matching sampler efficiently approximates the target distribution, significantly outperforming other sampling approaches.

The primary application of collaborate filtering (CF) is to recommend a small set of items to a user, which entails ranking. Most approaches, however, formulate the CF problem as rating prediction, overlooking the ranking perspective. In this work we present a method for collaborative ranking that leverages the strengths of the two main CF approaches, neighborhood- and model-based. Our novel method is highly efficient, with only seventeen parameters to optimize and a single hyperparameter to tune, and beats the state-of-the-art collaborative ranking methods. We also show that parameters learned on one dataset yield excellent results on a very different dataset, without any retraining.

Many areas of study, such as information retrieval, collaborative filtering, and social choice face the preference aggregation problem, in which multiple preferences over objects must be combined into a consensus ranking. Preferences over items can be expressed in a variety of forms, which makes the aggregation problem difficult. In this work we formulate a flexible probabilistic model over pairwise comparisons that can accommodate all these forms. Inference in the model is very fast, making it applicable to problems with hundreds of thousands of preferences. Experiments on benchmark datasets demonstrate superior performance to existing methods

We investigate the problem of learning to rank with document retrieval from the perspective of learning for multiple objective functions. We present solutions to two open problems in learning to rank: first, we show how multiple measures can be combined into a single graded measure that can be learned. This solves the problem of learning from a 'scorecard' of measures by making such scorecards comparable, and we show results where a standard web relevance measure (NDCG) is used for the top-tier measure, and a relevance measure derived from click data is used for the second-tier measure; the second-tier measure is shown to significantly improve while leaving the top-tier measure largely unchanged. Second, we note that the learning-to-rank problem can itself be viewed as changing as the ranking model learns: for example, early in learning, adjusting the rank of all documents can be advantageous, but later during training, it becomes more desirable to concentrate on correcting the top few documents for each query. We show how an analysis of these problems leads to an improved, iteration-dependent cost function that interpolates between a cost function that is more appropriate for early learning, with one that is more appropriate for late-stage learning. The approach results in a significant improvement in accuracy with the same size models. We investigate these ideas using LambdaMART, a state-of-the-art ranking algorithm.

Ranking a set of retrieved documents according to their relevance to a query is a popular problem in information retrieval. Methods that learn ranking functions are difficult to optimize, as ranking performance is typically judged by metrics that are not smooth. In this paper we propose a new listwise approach to learning to rank. Our method creates a conditional probability distribution over rankings assigned to documents for a given query, which permits gradient ascent optimization of the expected value of some performance measure. The rank probabilities take the form of a Boltzmann distribution, based on an energy function that depends on a scoring function composed of individual and pairwise potentials. Including pairwise potentials is a novel contribution, allowing the model to encode regularities in the relative scores of documents; existing models assign scores at test time based only on individual documents, with no pairwise constraints between documents. Experimental results on the LETOR3.0 data set show that our method out-performs existing learning approaches to ranking.

We present a system, ConEx, for monitoring query execution in a relational database management system. ConEx offers a unified view of query execution, providing continuous visual feedback on the progress of the query, and the status of operators in the query evaluation plan. It incorporates novel techniques to dynamically estimate important parameters affecting query progress efficiently. We describe the design and features of ConEx, and discuss its technology.