Large multimodal models (LMMs) have gained impressive performance due to their outstanding capability in various understanding tasks. However, these models still suffer from some fundamental limitations related to robustness and generalization due to the alignment and correlation between visual and textual features. In this paper, we introduce a simple but efficient learning mechanism for improving the robust alignment between visual and textual modalities by solving shuffling problems. In particular, the proposed approach can improve reasoning capability, visual understanding, and cross-modality alignment by introducing two new tasks: reconstructing the image order and the text order into the LMM's pre-training and fine-tuning phases. In addition, we propose a new directed-token approach to capture visual and textual knowledge, enabling the capability to reconstruct the correct order of visual inputs. Then, we introduce a new Image-to-Response Guided loss to further improve the visual understanding of the LMM in its responses. The proposed approach consistently achieves state-of-the-art (SoTA) performance compared with prior LMMs on academic task-oriented and instruction-following LMM benchmarks.

Time-frequency analysis (TFA) is a powerful method to exploit the hidden information of signals, including speech signals. Many techniques in this group were invented and developed to capture the most crucial stationary feature. However, human speech is not stable, and it contains some non-stationary elements. This work aims to design a new algorithm via the TFA technique to extract the trends and changes inside the speech signal in the time-frequency (TF) plane. We design a new algorithm to create a set of atoms for the signal transform, which can analyze the signal in many different view directions via Poly-Linear Chirplet Transform (PLCT). After processing the signal, the proposed method returns a multichannel output in which each channel results from a particular Linear Chirplet Transform (LCT). The feature then is combined with the MFCC feature to form the final representation. Although the size for speech representation rises, our extracted feature contains rich-meaning information to improve the recognition results compared to other features in gender recognition, dialect recognition, and speaker recognition.

The optimum multiuser detection (OMD) is a discrete (binary) optimization. The previously developed approaches often relax it by a semi-definite program (SDP) and then employ randomization for searching the optimal solution around the solution of this relaxed SDP. In this paper, we show the limited capacity of this SDP program, which at the end cannot give a better solution than the simple linear minimum mean square error detector (LMMSE). Our departure point is to express the problem as quadratic minimization over quadratic equality constraint (QMQE) or concave quadratic minimization over a box of continuous optimization (CQOB). The QMQE allows us to develop a nonsmooth optimization algorithm to locate the global optimal solution of OMD, while CQOB facilities effective confirmation of the solutions found by QMQE. Our intensive simulation clearly shows that the algorithm outperforms all previously developed algorithms while the computational burden is essentially reduced.

A new efficient method is developed for optimal maximum likelihood (ML) decoding of an arbitrary binary linear code based on data received from a Gaussian channel. The decoding algorithm is based on minimization of a difference of two monotonic objective functions subject to the 0-1 constraint of bit variables. The iterative process converges to the global optimal ML solution after a finite number of steps. The proposed algorithm's computational complexity depends on the input sequence length k which is much less than the codeword length n, especially for codes with small code rates. The viability of the developed method is verified through simulations on different coding schemes

In this paper, we develop a new linear matrix inequality (LMI) technique, which is practical for solutions of the general trigonometric semi-infinite linear constraint (TSIC) of competitive orders. Based on the new full LMI characterization for the convex hull of a trigonometric curve, it is shown that the semi-infinite optimization problem involving TSIC can be solved by an LMI optimization problem with additional variables of dimension just n, the order of the trigonometric curve. Our solution method is very robust which allows us to address almost all practical filter design problems. Unlike most previous works involving several complex mathematical tools, our derivation arguments are based on simple results of the convex analysis and some formal elementary transforms. Furthermore, many filter/filterbank design problems can be reformulated as the optimization of linear/convex quadratic objectives over the TSIC. Based on this reformulation, these problems can be equivalently reduced to LMI optimization problems with the minimal size. Our examples of designing up to 1200-tap filters verifies the viability of our formulation.

The advantage of IIR filters over FIR ones is that the former require a much lower order to obtain the desired response specifications. However, the existing deterministic techniques for IIR filter bank design based on heuristic usually lead to too high order IIR filters and thus cannot be practically used. In this paper, we propose new method to solve the low-order IIR filter bank design, which is based on linear matrix inequalities (LMI) optimization. Our focus is the QMF bank design, although other IIR filter related problems can be treated and solved in similar way.