All issues

Interpretability in recent deep learning models has become an epicenter of research particularly in sensitive domains such as healthcare, and finance. Concept bottleneck models have emerged as a promising approach for achieving transparency and interpretability by leveraging a set of humanunderstandable concepts as an intermediate representation before the prediction layer. However, manual concept annotation is discouraged due to the time and effort involved. Our work explores the potential of large language models (LLMs) for generating high-quality concept banks and proposes a multimodal evaluation metric to assess the quality of generated concepts. We investigate three key research questions: the ability of LLMs to generate concept banks comparable to existing knowledge bases like ConceptNet, the sufficiency of unimodal text-based semantic similarity for evaluating concept-class label associations, and the effectiveness of multimodal information in quantifying concept generation quality compared to unimodal concept-label semantic similarity. Our findings reveal that multimodal models outperform unimodal approaches in capturing concept-class label similarity. Furthermore, our generated concepts for the CIFAR-10 and CIFAR-100 datasets surpass those obtained from ConceptNet and the baseline comparison, demonstrating the standalone capability of LLMs in generating highquality concepts. Being able to automatically generate and evaluate high-quality concepts will enable researchers to quickly adapt and iterate to a newer dataset with little to no effort before they can feed that into concept bottleneck models.

Mathematical models of gas dynamics and its computational industry, in our opinion, are far from perfect. We will look at this problem from the point of view of a clear probabilistic micro-model of a gas from hard spheres, relying on both the theory of random processes and the classical kinetic theory in terms of densities of distribution functions in phase space, namely, we will first construct a system of nonlinear stochastic differential equations (SDE), and then a generalized random and nonrandom integro-differential Boltzmann equation taking into account correlations and fluctuations. The key feature of the initial model is the random nature of the intensity of the jump measure and its dependence on the process itself.

Briefly recall the transition to increasingly coarse meso-macro approximations in accordance with a decrease in the dimensionalization parameter, the Knudsen number. We obtain stochastic and non-random equations, first in phase space (meso-model in terms of the Wiener — measure SDE and the Kolmogorov – Fokker – Planck equations), and then — in coordinate space (macro-equations that differ from the Navier – Stokes system of equations and quasi-gas dynamics systems). The main difference of this derivation is a more accurate averaging by velocity due to the analytical solution of stochastic differential equations with respect to the Wiener measure, in the form of which an intermediate meso-model in phase space is presented. This approach differs significantly from the traditional one, which uses not the random process itself, but its distribution function. The emphasis is placed on the transparency of assumptions during the transition from one level of detail to another, and not on numerical experiments, which contain additional approximation errors.

The theoretical power of the microscopic representation of macroscopic phenomena is also important as an ideological support for particle methods alternative to difference and finite element methods.

The method for mapping of intermediate representations (IR) set of C, C++ programs to vector embedding space is considered to create an empirical estimation framework for static performance prediction using LLVM compiler infrastructure. The usage of embeddings makes programs easier to compare due to avoiding Control Flow Graphs (CFG) and Data Flow Graphs (DFG) direct comparison. This method is based on transformation series of the initial IR such as: instrumentation — injection of artificial instructions in an instrumentation compiler’s pass depending on load offset delta in the current instruction compared to the previous one, mapping of instrumented IR into multidimensional vector with IR2Vec and dimension reduction with t-SNE (t-distributed stochastic neighbor embedding) method. The D1 cache miss ratio measured with perf stat tool is considered as performance metric. A heuristic criterion of programs having more or less cache miss ratio is given. This criterion is based on embeddings of programs in 2D-space. The instrumentation compiler’s pass developed in this work is described: how it generates and injects artificial instructions into IR within the used memory model. The software pipeline that implements the performance estimation based on LLVM compiler infrastructure is given. Computational experiments are performed on synthetic tests which are the sets of programs with the same CFGs but with different sequences of offsets used when accessing the one-dimensional array of a given size. The correlation coefficient between performance metric and distance to the worst program’s embedding is measured and proved to be negative regardless of t-SNE initialization. This fact proves the heuristic criterion to be true. The process of such synthetic tests generation is also considered. Moreover, the variety of performance metric in programs set in such a test is proposed as a metric to be improved with exploration of more tests generators.